Attitudes Toward Emotions

Home , Affect measures, Attitude (psychology)

Eddie Harmon-Jones and Cindy Harmon-Jones David M. Amodio Texas A&M University New York University

Philip A. Gable University of Alabama

The present work outlines a theory of attitudes toward emotions, provides a measure of attitudes toward emotions, and then tests several predictions concerning relationships between attitudes toward specific emotions and emotional situation selection, emotional traits, emotional reactivity, and emotion regulation. The present conceptualization of individual differences in attitudes toward emotions focuses on specific emotions and presents data indicating that 5 emotions (anger, sadness, joy, fear, and disgust) load on 5 separate attitude factors (Study 1). Attitudes toward emotions predicted emotional situation selection (Study 2). Moreover, attitudes toward approach emotions (e.g., anger, joy) correlated directly with the associated trait emotions, whereas attitudes toward withdrawal emotions (fear, disgust) correlated inversely with associated trait emotions (Study 3). Similar results occurred when attitudes toward emotions were used to predict state emotional reactivity (Study 4). Finally, attitudes toward emotions predicted specific forms of emotion regulation (Study 5).

Keywords: discrete emotions, approach motivation, withdrawal motivation, emotion regulation, behavioral activation system (BAS)

Emotions pervade subjective experience (Izard, 2009), and al- theory by assessing relationships between attitudes toward specific though often perceived as a single subjective state, emotional emotions and emotional traits, emotional reactivity, emotional experience is likely composed of many different elements. Tom- situation selection, and emotion regulation. kins (1962, 1963) and others (Ellsworth, 1994; Izard, 1971) suggested that the evaluation of an emotion is part of the experience of emotion. Although the importance of attitudes toward emotions Review of Past Research on Attitudes Toward was recognized almost half a century ago, the empirical study of Emotions them has been rather limited. Most research on this topic has examined cultural differences in attitudes toward emotions and not Izard (1971) considered attitudes toward emotions as part of the examined how these attitudes relate to other emotion-related vari- emotional experience. Most of the questions on his scale asked ables. participants to indicate one of the fundamental emotions as the The present research takes a new perspective on attitudes toward answer. For example, one question asked, “Which emotion do you emotions. By attitudes, we mean subjective evaluations that range most prefer to experience?” Other questions concerned under- from good to bad that are represented in memory; this definition is standing emotions and frequency of experience. Thus, this ques- consistent with social psychological definitions of attitudes (Olson tionnaire measured concepts that are not considered attitudes in & Zanna, 1993). In this research, we examine the relationship contemporary attitudes research, and its attitude questions assessed between attitudes toward the private experience of emotions and one-word (emotion) responses, making it difficult to conduct linear different aspects of emotional experience and responding. We analyses. Subsequent research on attitudes toward emotion used outline a theory of attitudes toward emotions, present a new similar measurements (Sommers, 1984). Moreover, this research method for their measurement, and test specific aspects of the focused on cultural differences. More recently, Eid and Diener (2001) found that different cultures have different norms for the expression and experience of This article was published Online First August 15, 2011. various emotions. In their work, participants indicated “how ap- Eddie Harmon-Jones and Cindy Harmon-Jones, Department of Psychol- propriate or desirable it is to experience certain emotions,” and ogy, Texas A&M University; David M. Amodio, Department of Psychol- responses to several one-item questions for each emotion were ogy, New York University; Philip A. Gable, Department of Psychology, assessed. They found a number of differences in the desirability of University of Alabama. different emotions across cultures. For example, individuals in This research was funded in part by grants from the National Science individualistic cultures are more likely to believe that pride is a Foundation (BCS 0643348 and BCS 0921565) awarded to Eddie Harmon- Jones. desirable emotion, whereas individuals in collectivist cultures are Correspondence concerning this article should be addressed to Eddie more likely to believe that guilt is a desirable emotion. Eid and Harmon-Jones, Department of Psychology, Texas A&M University, 4235 Diener noted that their results cast doubt on the classification of TAMU, College Station, TX 77845. E-mail: [email protected] emotions as positive or negative, because an emotion that is

1332 ATTITUDES TOWARD EMOTIONS 1333 considered negative in one culture may be considered positive in Also, our conceptualization emphasizes a specific emotion per- another culture. spective that is interested in differences between emotions such as Tsai, Knutson, and Fung (2006) proposed affect valuation the- fear and anger—emotions that are considered similar in the ideal ory. It proposes that (a) ideal affect, the states that people value affect framework, given that both are high in negative valence and and would ideally like to feel, differs from actual affect, the states arousal. In addition, the present work focuses on attitudes toward that people actually feel; and that (b) cultural factors shape ideal emotions rather than subjective norms about emotions, as in Eid affect more than actual affect, whereas temperament shapes actual and Diener (2001). Moreover, it goes beyond past work that has affect more than ideal affect. Affect valuation theory proposes that focused primarily on cultural differences (as in Eid & Diener, ideal affect refers to a goal and that actual affect refers to a 2001; Tsai et al., 2006), by examining individual differences in response. Whereas ideal affect requires some understanding of attitudes toward emotions within a culture and how they predict different affective states and their contingencies, actual affect does other important emotion-related variables. not. Finally, because this theorizing assumes that most people want We believe that both discrete and dimensional approaches are to feel good, ideal affect should primarily involve different posi- important, and their importance for a given research question tive states, whereas actual affect should instead involve the entire likely depends on the level of analysis of the emotional process spectrum of affective states. Moreover, variation in ideal affect is under consideration (e.g., feeling, physiology, behavioral expres- predicted to be primarily due to cultural ideas and practices. sion). Several dimensions likely underlie discrete emotions. Va- Building on a Valence ϫ Arousal model of affect, Tsai and lence and arousal dimensions have received much research atten- colleagues assessed ideal values for positive versus negative va- tion, but they do not fully capture emotional space, as research has lence as a function of low versus high arousal. European Ameri- revealed important distinctions between fear and anger (Carver & cans were found to value high-arousal positive states more and Harmon-Jones, 2009; E. Harmon-Jones & Peterson, 2008; Lerner low-arousal positive states less than Chinese individuals (Tsai, & Keltner, 2001; Mendes, Major, McCoy, & Blascovich, 2008), 2007). two emotions that occupy identical space in a Valence ϫ Arousal model. Adding the dimension of motivational direction—approach A Theory of Attitudes Toward Emotions versus withdrawal—assists us in understanding some differences between anger and fear, as anger is often approach motivating The present work on attitudes toward emotions extends past (Carver & Harmon-Jones, 2009), whereas fear is often withdrawal work. We conceive of emotional experience as involving many motivating (K. A. Buss et al., 2003). However, the addition of elements. Individuals report differences in feelings of arousal and motivational direction does not assist in separating fear from valence. Individuals also have subjective experiences of the action disgust. Indeed, other dimensions beyond valence, arousal, and tendencies of emotions (Frijda, 1986). And there are likely many motivational direction are important in characterizing emotions other aspects that compose emotional experience. We posit that (Frijda, & Tcherkassof, 1997). part of emotional experience may be the subjective evaluation of In agreement with past research showing many cultural similar- the private emotional experience or the attitude toward emotion. ities in attitudes toward emotions, our theory predicts that the An individual in a bout of anger may feel aroused, may feel range and quality of individual differences in attitudes toward negative about the perceived cause of the anger, may feel tenden- emotions are likely restricted by evolutionary, universal mecha- cies toward screaming, and may really dislike the experience of nisms. In other words, attitudes toward positive emotions, such as anger. joy, are likely more positive than attitudes toward negative emo- Our theory begins by predicting that attitudes toward emotions tions, such as anger, fear, sadness, and disgust. For example, are organized in a discrete emotions framework; that is, individuals organisms that lacked a sufficiently negative experience of fear should have distinct attitudes toward discrete emotions, and not would lack the motivation to flee threats. simply attitudes toward valence and arousal dimensions. It further In addition, the present work should aid in defining the subjec- predicts that attitudes toward discrete emotions should predict tive valence of the emotion. In discussing definitions of emotional emotion situation selection, such that disliking of particular emo- valence, Lazarus (1991) noted that emotions can be regarded as tions should be associated with avoidance of situations that would positive or negative on the basis of (a) the conditions that evoked evoke those emotions. The theory also predicts that attitudes the emotion, (b) the emotion’s adaptive consequences, or (c) the toward discrete emotions will relate to trait and state emotions but emotion’s subjective feel. Attitudes toward emotions concern this that the direction of those relationships will depend on the ap- latter definition. For example, negative emotions such as anger or proach versus avoidance quality of the emotion. That is, approach- fear may vary in how negatively they are experienced by individ- oriented emotions (e.g., joy, anger) should be liked the more they uals, depending on the degree to which an individual likes or are experienced, whereas avoidance-oriented emotions (e.g., fear, dislikes the subjective experience associated with the particular disgust) should be disliked the more they are experienced. Finally, emotion. Indeed, discrete emotion theorist Paul Ekman (2003, p. the theory predicts that attitudes toward emotions will relate to the 190) suggested that the liking or enjoyment of emotions may down-regulation of emotion, such that, for example, more dislike underlie which emotions are thought of as positive or negative: of fear should relate to more avoidance of fear particularly once “Just as there is a set of distinctive emotions that we usually don’t fear is evoked. The reasoning underlying these theoretically based enjoy feeling, there is a set of distinctive emotions that we do enjoy predictions is expanded upon below. feeling.” The present conceptualization is similar to past work. However, By incorporating the concept of evaluations or attitudes toward unlike past work, the present theory focuses on liking for specific emotions into research and theory on emotion processing and emotions, not ideal dimensional states as in affect valuation theory. reactions, we should be better positioned to understand and predict 1334 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE emotional behaviors. Below, we advance some specific predictions differs from the prediction advanced by affect valuation theory, that follow from considering attitudes toward specific emotions as which predicts that variation in ideal affect is primarily due to vital parts of emotional experience. cultural ideas and practices. Building on work illustrating the importance of the dimension of Prediction 1: Attitudes Toward Emotions Organize in motivational direction in emotions (C. Harmon-Jones, Schmeichel, a Discrete Emotions Framework Mennitt, & Harmon-Jones, 2011; E. Harmon-Jones, 2003), we predicted relationships of different directions (direct vs. inverse) We expected that attitudes toward emotions would organize in a between trait emotions and attitudes toward those emotions.1 For discrete emotions framework. In these initial tests of our model, approach-oriented emotions, such as anger and joy (Carver & we focused on attitudes toward anger, fear, disgust, sadness, and Harmon-Jones, 2009), we predicted that higher trait levels of the joy. These emotions were chosen because they are the most fre- emotion would relate to greater liking of the emotion. In contrast, quently examined emotions in past research (Ekman, 2003). An- for withdrawal-oriented emotions, such as fear and disgust, we ger, fear, disgust, and sadness are classified as negative emotions, predicted that higher trait levels of the emotion would relate to and joy is classified as a positive emotion (Lazarus, 1991). In addition to these valence differences, these emotions differ in greater disliking of the emotion. arousal level and motivational direction. Regarding arousal Approach- and withdrawal-oriented emotions may relate in dif- level, anger, fear, disgust, and joy are typically considered to be ferent directions to attitudes toward emotions, because both emo- more arousing than sadness (Lang, Bradley, & Cuthbert, 1990). tions and attitudes toward emotions assist in guiding and energiz- Regarding motivational direction, anger and joy are typically con- ing behavior. They both assist in motivating behavior that sidered to be moderate to high in approach motivation. In contrast, corresponds to the emotion itself. Fear and the dislike of fear both fear and disgust are considered to be moderate to high in with- promote avoidance motivation. When individuals are afraid, they drawal motivation (Carver & Harmon-Jones, 2009). Sadness is are motivated to avoid the situation causing the fear. Similarly, if often characterized as a low approach emotion (Carver, 2004; they dislike the experience of fear, they should be even more Henriques & Davidson, 2000; Higgins, Shah, & Friedman, 1997), motivated to avoid the fear and consequently the situation causing but at other times, it is characterized as a withdrawal emotion the fear. Anger, on the other hand, promotes approach motivation, (K. A. Buss et al., 2003). Consistent with the idea that sadness is the urge to act outwardly on the environment (Carver & Harmon- associated with approach, sadness occurs following failure when Jones, 2009; Ford et al., 2010). When individuals are angry, they individuals are in an approach motivational orientation (Carver, are motivated to approach the situation or anger-eliciting target. If 2004; Higgins et al., 1997). they like the experience of anger, they should be even more motivated to approach the feeling of anger as well as the anger- Prediction 2: Attitudes Toward Emotions Influence eliciting situation or target. Emotional Situation Selection This hypothesis that subjective emotional intensity and attitudes toward emotion should be directly correlated for approach emo- Attitudes toward emotions are predicted to influence the emotions and inversely correlated for withdrawal emotions, because tional situations individuals select. We predicted that the majority both emotional intensity and attitudes assist in motivating behavior of individuals would like joy and dislike fear, disgust, anger, and that corresponds to the emotion itself, may be similar to findings sadness; however, the degree to which individuals like and dislike these emotions will vary. Just as persons with greater dislike of in attitudes/emotions research. For instance, approach-related ac- sushi are more likely to avoid situations where sushi might be tions are typically associated with more liking, whereas encountered (compared to individuals with a slight dislike for withdrawal-related actions are typically associated with more dis- sushi), persons who greatly dislike fear should be more likely to liking (Cacioppo, Priester, & Berntson, 1993). This work exam- avoid situations that arouse fear (compared to persons with lesser ined the effects of approach versus withdrawal actions on attitudes dislike for fear), and persons who greatly dislike disgust should be toward neutral or ambiguous external stimuli. But the process more likely to avoid situations that arouse disgust. Similarly, could be similar for internal stimuli such as emotions. Approach- persons with greater dislike for anger should be more likely to related actions associated with anger may lead to more liking of avoid angering situations, and persons with greater dislike for the internal experience of anger, whereas the withdrawal-related sadness should be more likely to avoid situations that arouse actions associated with fear may lead to more disliking of the sadness. Finally, persons with greater liking for joy should be more internal experience of fear. Past research has revealed that trait likely to approach situations that arouse joy than persons with anger is directly related to attitudes toward anger (E. Harmon- lower liking for joy. Jones, 2004), but no research has examined relations between other attitudes toward emotions and other trait emotions. Prediction 3: Individual Differences in Attitudes Toward Emotions Are Related to Trait Emotions We predicted that attitudes toward specific emotions would be related to those specific emotions at the trait level. We assumed 1 When describing the direction of correlations, we use the terms direct that the experience of emotions would influence the attitudes and inverse to denote positive and negative correlations, respectively. We individuals have toward these emotions (and vice versa). This do this to avoid confusion when describing correlations involving positive prediction that attitudes toward emotions relate to trait emotions and negative emotions or attitudes. ATTITUDES TOWARD EMOTIONS 1335

Prediction 4: Attitudes Toward Emotions Influence plications of attitudes toward emotion for emotion regulation Emotional Reactions (Study 5).

We predicted that attitudes toward emotions would influence Study 1 emotional reactions to emotionally evocative situations. As with the relations with trait emotions, the directions of these correla- Study 1 was conducted to test the factor structure and reliability tions between attitudes toward emotions and emotional reactions of the ATE. It was also conducted to examine intercorrelations were predicted to vary as a function of whether the emotions are among ATE subscales and to test mean-level differences between approach or withdrawal motivated. That is, liking of approach- these subscales. oriented emotions, such as joy and anger, should be associated with more intense joy and anger reactions in situations evoking joy Method or anger, respectively. In contrast, liking of withdrawal-oriented Participants were from two samples of college students who emotions, such as fear and disgust, should be associated with less participated in exchange for extra credit in their introductory intense fear and disgust reactions in situations evoking fear or psychology course. The first sample comprised 1,523 participants disgust, respectively. and the second sample comprised 1,041 participants.2 Combining Predictions 2, 3, and 4, the following emerges: If Both samples completed the Attitudes Toward Emotions Scale people dislike fear greatly, then they are predicted to experience (ATE; see Table 1) in group sessions. Participants were given the more fear when exposed to a fearful situation and to be more following instructions: “There are no right or wrong answers for fearful generally. Also, individuals who greatly dislike fear will the following questions. Please answer honestly based on how you attempt to avoid fearful situations. One may then wonder why feel. Thank you very much for your participation. Please answer individuals with greater dislike for fear would experience more each question using the scale provided below.” Ratings were made fear. That is, if these people avoid fear-evoking situations, then on the following scale: 1 ϭ rarely/never,2ϭ occasionally,3ϭ they should have no fear. We propose that it is almost impossible sometimes,4ϭ often,5ϭ almost always/always. Items for the to successfully avoid all fear-producing stimuli. Fear-evoking ATE were generated on the basis of their face validity regarding stimuli—heights, angry dogs, symptoms of deadly diseases, nat- the underlying construct. For example, attitudes toward anger were ural disasters, insects, and so forth—may be encountered at any measured by asking participants to rate the frequency with which time, even if one avoids voluntarily encountered fear-evoking they liked various aspects of angry experiences. The current Atti- situations such as scary movies or skydiving. Thus, individuals tudes Toward Anger measure contained a subset of items from the who greatly dislike fear may attempt to avoid fearful situations but previously published Attitudes Toward Anger Scale (E. Harmon- are unlikely to be completely successful, and consequently they Jones, 2004). continue to experience much fear. Results Prediction 5: Attitudes Toward Emotions Influence Emotion Regulation Exploratory factor analysis. In light of concerns over the use of only one type of factor analysis for evaluating personality Our conceptualization of attitudes toward emotions has impli- scales (Hopwood & Donnellan, 2010), we conducted exploratory cations for emotion regulation processes. Individuals often attempt and confirmatory factor analyses. An exploratory principal com- to reduce their experience of negative emotions. Perhaps one ponents analysis with varimax normalized rotation was conducted reason they are motivated to “down-regulate” their negative emo- on the first sample. It revealed five components with eigenvalues tions is because they dislike the negative emotion. However, to our greater than 1.0. The scree plot analysis also supported a five- knowledge, this assumption has not been tested empirically. We factor solution. For each component, items from a sole ATE Ͼ predicted that attitudes toward negative emotions such as fear subscale loaded on that particular component ( .40), and no should predict attempts to down-regulate that emotion. Specifi- cross-loadings resulted. Individual loadings are shown in Table 1. cally, individuals who strongly dislike a particular emotion should Confirmatory factor analysis. Our next goal was to deter- be more motivated to down-regulate that emotion compared with mine whether the ATE was composed of attitudes toward five individuals who dislike that emotion less. different emotions, using confirmatory factor analysis, which per- mits a statistical test of the fit of the theoretical model against the actual data. We first tested the fit to our five-factor model, which The Present Research has attitudes toward each of the five specific emotions on separate factors. A second confirmatory factor analysis tested a two-factor The present research was designed to test the predictions of our valence-based model that has one factor for positive emotions and model outlined above. To this end, we conducted five studies to one factor for negative emotions. A third confirmatory factor establish the validity and utility of the attitudes toward emotions analysis tested a three-factor Valence ϫ Arousal model; it has one (ATE) construct. In Study 1, we conducted tests of the psycho- factor for high-arousal negative emotions (anger, disgust, fear), metric properties of the ATE, examining its factor structure and another factor for low-arousal negative emotions (sadness), and internal consistency. Then, we tested whether the ATEs would another factor for positive emotions (joy; low- to high-arousal predict the choice to view specific emotional stimuli (Study 2). We then examined the ATE’s relationships with affective traits (Study 3) and emotional reactivity (Study 4). Finally, we examined im- 2 Gender information was not collected in Studies 1 or 2. 1336 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE

Table 1 Scale Items and Loadings of Items on Five Principal Components

Subscale and item Loading

Attitude Toward Anger I like the feeling of increased energy I get from expressing my anger. .80 I like the feeling of power I get from expressing my anger. .85 I like it when I feel like yelling at someone. .77 I dislike how it feels when I am angry. Ϫ.50 I like how it feels when I am furious. .67 Attitude Toward Joy I do not really enjoy the moments in my life when I am happy. Ϫ.64 I like experiencing joy. .79 I prefer to hang around with people who make me happy. .73 I really like feeling happy. .80 I like conversations that make me feel happy. .79 Attitude Toward Sadness I like it when movies make me feel sad, the sadder the better. .77 If someone describes a movie as a real “tear jerker,” I am sure to avoid it because I don’t like feeling sad. Ϫ.55 I like thinking about sad things. .56 I find myself reading sad books. .67 I can enjoy a conversation even though it makes me sad. .59 If a book, movie, or TV show makes me cry, I know I’ve really enjoyed it. .72 Attitude Toward Disgust If I hear something disgusting, I will listen to it again on purpose. .74 If I see something disgusting, I will look at it again on purpose. .77 I do not enjoy doing things that I find disgusting. Ϫ.44 If I smell something disgusting, I will smell it again on purpose. .68 I like doing things that I find disgusting. .70 If I feel something disgusting, I will feel it again on purpose. .75 Attitude Toward Fear I like to do things that scare me. .76 I do things just because they scare me. .75 I like being scared. .79 I seek out things that scare me. .83 I dislike being scared. Ϫ.79 I dislike doing things that scare me. Ϫ.74

Note. Items are from the Attitudes Toward Emotions Scale. Loadings are from the principal components analysis. positive emotions were not included because of our five-factor was not acceptable (.0675, 90% CI [.0646, .0704]). The Jöres- model prediction). A fourth and final confirmatory factor analysis kog GFI and AGFI were inferior to those for the five-factor was conducted to test a one-factor model that has attitudes for all model (.86 and .84, respectively). five emotions loading on one global factor. Comparison of the fit Comparison of the difference in chi-square between the five- of these four models was also performed. factor model and the two-factor model revealed that the five-factor Fit to five-factor model. To assess the goodness of fit of the model was a better fit than the two-factor model, difference in five-factor solution, the 28 items were submitted to a confirmatory ␹2(9) ϭ 789.32, p Ͻ .001. These results suggest that a five specific factor analysis with the initial model based on the prediction that five emotion model fit the data better than a two valence (positive– emotion-specific factors existed. All factors were allowed to covary, negative) emotion model. and the method of estimation was generalized least squares. Fit to three-factor model. We next tested the fit of a three- Indices of fit indicated that the five-factor model was accept- factor Valence ϫ Arousal model, in which one factor included able. That is, the Steiger-Lind root-mean-square error of approx- high-arousal negative emotion, one factor included the low-arousal imation (RMSEA) for the five-factor model was acceptable (.0497, negative emotion, and the other factor included the positive emo- 90% CI [.0467, .0527]). The Jöreskog goodness-of-fit index (GFI) and the adjusted goodness-of-fit index (AGFI) were also accept- tion. The Steiger-Lind RMSEA for the three-factor model was not able (.92 and .90, respectively). These results suggest that partic- acceptable (.061, 90% CI [.064, .070]). The Jöreskog GFI and ipants hold distinct attitudes toward each of the five emotions AGFI were inferior to those for the five-factor model (.88 and .85, included in the ATE. respectively). Fit to two-factor model. We next tested the fit of a two- Comparison of the difference in chi-square between the five- factor “valence” model, in which one factor included all of the factor model and the three-factor model revealed that the five- negative emotions and the other factor included the positive factor model was a better fit than the three-factor model, difference emotion. The Steiger-Lind RMSEA for the two-factor model in ␹2(7) ϭ 603.79, p Ͻ .001. These results suggest that a five ATTITUDES TOWARD EMOTIONS 1337 specific emotion model fit the data better than a Valence ϫ individuals had the most positive attitudes toward joy (Ms and SDs Arousal emotion model. for Samples 1 and 2 ϭ 4.66 [0.56] and 4.72 [0.57]) compared with Fit to one-factor model. This confirmatory factor analysis the negative emotions (ps Ͻ .001). Within the negative emotions, tested whether a single attitude toward the experience of all emo- sadness (Ms and SDs for Samples 1 and 2 ϭ 2.69 [0.72] and 2.58 tions existed. The Steiger-Lind RMSEA for the one-factor model [0.69]) was evaluated more positively than fear (p Ͻ .001; Ms and was not acceptable (.0715, 90% CI [.0687, .0744]). The Jöreskog SDs for Samples 1 and 2 ϭ 2.44 [0.96] and 2.33 [0.89]). Fear was GFI and AGFI were inferior to those for the five-factor model (.85 evaluated more positively than anger (ps Ͻ .001; Ms and SDs for and .82, respectively). Samples 1 and 2 ϭ 2.01 [0.83] and 1.89 [0.77]) and disgust (Ms Comparison of the difference in chi-square between the five- and SDs for Samples 1 and 2 ϭ 1.88 [0.75] and 1.83 [0.74]). factor model and the one-factor model revealed that the five-factor Finally, anger and disgust were not evaluated differently (p Ͼ .10). model was a better fit than the one-factor model, difference in ␹2(10) ϭ 998.25, p Ͻ .001. These results suggest that the five Discussion specific emotion model fit the data better than a model with a single factor for attitudes toward all emotions. Study 1 revealed that the ATE Scale had five factors, represent- Internal consistency. To determine the internal consistency ing distinct attitude toward emotion subscales—anger, disgust, of the ATE subscales, Cronbach’s alpha was computed for each of fear, joy, and sadness. Results did not support a Valence ϫ the five subscales identified by the five-factor confirmatory factor Arousal dimensional model. The subscales are internally consis- analysis. Satisfactory alpha coefficients were observed for each tent and reliable across time. Moreover, attitudes toward negative subscale, indicating that each was internally reliable. Cronbach’s emotions were directly and moderately correlated with each other. alphas for Samples 1 and 2, respectively, were as follows: for In contrast, attitude toward joy was inversely correlated with Attitude Toward Anger, .76 and .82; for Attitude Toward Disgust, attitudes toward the negative emotions. As expected, individuals .82 and .82; for Attitude Toward Fear, .88 and .88; for Attitude held a more positive attitude toward joy than toward the negative Toward Joy, .70 and .84; and for Attitude Toward Sadness, .67 and emotions. Attitudes toward negative emotions were generally neg- .70. ative, but they differed: Sadness was evaluated more positively Test–retest reliability. A random sample of 85 participants than fear, fear was evaluated more positively than anger and who completed the ATE at the beginning of the semester was disgust, and anger and disgust were not evaluated differently. administered the ATE again 10 weeks later. The test–retest reliability of the negative emotion scales of the ATE was adequate Study 2 (Anger, r ϭ .63; Disgust, r ϭ .67; Fear, r ϭ .82; Sadness, r ϭ .73). The test–retest reliability of the Joy scale of the ATE was not Consistent with predictions, Study 1 revealed that attitudes adequate (r ϭ .33), likely because of the restricted range (the range toward specific emotions load on separate factors and that joy is of joy was half the range of other emotions) and lack of variance evaluated more positively than the negative emotions. In Study 2, in the attitude toward joy (Joy SD ϭ 0.34, whereas the next lowest we examined whether ATE scores predicted participants’ prefer- SD was 0.64 for Sadness). That is, almost all participants reported ence for different types of emotional stimuli. We predicted that highly positive ratings of attitudes toward joy. individual differences in attitudes toward emotions would relate to Correlations among scales. As expected, the scales corre- participants’ interest in viewing attitude-relevant emotional stim- lated with each other, as displayed in Table 2. In general, attitudes uli. That is, individuals who have relatively more positive attitudes toward negative emotions were directly and moderately correlated toward a particular emotion should express more interest in view- with each other. Attitude toward joy was inversely correlated with ing photographs evoking that specific emotional state. In addition attitudes toward the negative emotions. to testing this primary hypothesis, we also examined whether Differences in attitudes toward emotions. To assess the assessments of trait emotional experience could account for the differences in attitudes toward the emotions, a repeated-measures effects of ATE on interest in viewing relevant emotional stimuli, to analysis of variance (ANOVA) was performed. It revealed that test the alternative explanation that trait emotional experience accounts for the relationship of ATE and interest in viewing attitude-relevant emotional stimuli. Table 2 Correlations Among ATE Subscales, Study 1 Method

ATE subscale ATA ATD ATF ATS ATJ Participants and procedure. Two hundred and two introductory psychology students completed the ATE and the Differ- ATA — ential Emotions Scale (DES–IV; Izard, Libero, Putnam, & Haynes, ATD .43 (.40) — 1993) in exchange for extra credit in their course. The DES–IV is ATF .34 (.38) .45 (.45) — ATS .17 (.14) .12 (.18) .12 (.10) — a 36-item scale comprising 12 subscales for each trait emotion. ATJ Ϫ.34 (Ϫ.30) Ϫ.30 (Ϫ.28) Ϫ.18 (Ϫ.15) Ϫ.13 (Ϫ.08) — Participants were asked to rate the extent to which they experienced various emotions and emotion-related responses on a scale Note. All correlations are significant p Ͻ .01. Correlations displayed ranging from 1 (rarely or never)to5(very often). inside parentheses are from Sample 1, and those outside parentheses are from Sample 2. ATE ϭ Attitudes Toward Emotion Scale; ATA ϭ Attitude Between 4 and 12 weeks later, they were invited to an individual Toward Anger; ATD ϭ Attitude Toward Disgust; ATF ϭ Attitude Toward testing session in which they were told that many pretesting studies Fear; ATS ϭ Attitude Toward Sadness; ATJ ϭ Attitude Toward Joy. were being conducted. They learned that the first of these pretests 1338 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE involved looking at five separate pieces of paper, each displaying evoking, sadness-evoking, and fear-evoking photographs but more a set of five color pictures taken from the International Affective interest in viewing joy-evoking photographs. To explore whether Picture System (IAPS; Lang, Bradley, & Cuthbert, 1999). Each specific ATE subscales predicted more interest in viewing photo- page contained pictures associated with one of five different graphs from the specific corresponding emotion type over the emotion-related categories (fear, disgust, sadness, and joy, plus a other emotion types, regression analyses were conducted in which page of affectively neutral pictures). scores on each ATE subscale were entered simultaneously as Prior to giving participants a notebook containing the five sets predictor variables and interest in viewing one of each set of of pictures, the experimenter told participants that they should emotion photographs was the criterion variable.3 Interest in view- view the pictures before completing questionnaires about them. ing neutral pictures was also included as a covariate to remove The experimenter then explained that they should select one of the variance due to individual differences in use of the response scales five picture sets that they would like to view later in the testing and general interest in viewing photographs. session, in which the participant would view pictures related to For interest in viewing joy-evoking photographs, the overall only one of the five pages (ostensibly due to time constraints). In regression equation was significant, R2 ϭ .14, F(6, 190) ϭ 4.98, all, they were told they would view and rate 20 pictures of the p Ͻ .001. Only interest in viewing neutral photographs (␤ϭ.27, selected emotion category. Before participants viewed and rated p ϭ .0001) and more positive attitudes toward joy (␤ϭ.19, p ϭ the 20 “test” pictures, the experimenter reentered the participant’s .006) predicted more interest in viewing joy-evoking photographs. cubicle and began a funneled debriefing, in which the participant For interest in viewing disgust-evoking photographs, the overall was carefully probed for suspicion and then informed of the regression equation was significant, R2 ϭ .18, F(6, 190) ϭ 6.60, hypotheses. p Ͻ .001. Only more negative attitudes toward disgust predicted Materials. The five separate pieces of paper each displayed less interest in viewing disgust-evoking photos (␤ϭ.34, p ϭ pictures from one emotion category. These pictures were found to .00002). evoke the intended emotions in pretesting (see also results from For interest in viewing fear-evoking photos, the overall regres- current Study 4). The fear pictures consisted of an open-mouthed sion equation was significant, R2 ϭ .14, F(6, 190) ϭ 5.14, p Ͻ snake (IAPS #1050), a handgun pointed at the viewer (#6230), a .001. In this analysis, more negative attitudes toward fear and spider on a web (#1220), a man holding a knife to a woman’s neck toward disgust independently predicted less interest in viewing (#6350), and a growling dog (#1300). The disgust pictures con- fear-evoking photos (␤ϭ.19, p ϭ .02, and ␤ϭ.20, p ϭ .01, sisted of a dirty toilet (#9300), a pizza with cockroaches (#7380), respectively). a decaying animal corpse (#9570), a severely injured human face For interest in viewing sadness-evoking photos, the overall (#3060), and a burnt human torso (#3110). The sadness pictures regression equation was significant, R2 ϭ .11, F(6, 190) ϭ 5.18, displayed an elderly couple in the hospital (#2205), a young p Ͻ .001. In this analysis, more negative attitudes toward sadness malnourished African crying (#2800), an elderly person dressed in and toward disgust predicted less interest in viewing sadness- black standing in the rain (#9210), a large group of individuals evoking photos (␤ϭ.17, p ϭ .02, and ␤ϭ.29, p ϭ .0004, leaving a wrecked airplane (#9050), and two adults standing over respectively). a grave in a sad posture (#9220). The joy pictures displayed cute For interest in viewing neutral photos, the overall regression kittens (#1463), a laughing baby (#2040), a desirable ice cream equation was not significant, R2 ϭ .05, F(5, 191) ϭ 1.87, p ϭ .10. cone (#7330), laughing chimpanzees (#1811), and individuals sky- In this analysis, more negative attitudes toward sadness predicted diving (#5621). The neutral pictures displayed a truck (#7130), an more interest in viewing neutral photos (␤ϭ–.15, p ϭ .04). This electrical outlet (#6150), a highway (#7560), a dinner plate latter result may be consistent with studies showing depressed (#7233), and a stereo speaker (#7185). No pictures designed to individuals have negative cognitive/emotional biases toward neu- evoke anger were included because the IAPS does not contain tral stimuli (e.g., Schwartz, Fair, Salt, Mandel, & Klerman, 1976). many pictures that evoke primarily anger in all viewers. Controlling for trait emotional experience. To assess The questionnaire about the pictures asked participants to rate whether the above relationships would continue to exist when how much they “would like to view a set of 20 pictures similar to controlling for trait emotional experience, simultaneous regression those presented on [a particular sheet number]?” The five sheets analyses were conducted in which the predictors included scores were numbered A1 to A5 or B1 to B5 or C1 to C5, to reflect three for (a) attitude toward a specific emotion; (b) a specific, relevant random orders in which picture types were presented. Ratings were trait emotion; and (c) interest in viewing neutral pictures. The made on 1 (not at all)to9(very much) scales. criterion was participants’ interest in viewing the emotion-relevant pictures. For instance, in a regression examining participants’ Results interest in viewing fear stimuli, the predictors included attitude toward fear, trait fear, and interest in viewing neutral stimuli. It is As shown in the correlation table (see Table 3), each ATE important to note that correlations between DES emotions and subscale predicted interest in viewing its type of emotion photo- attitudes toward emotions were not strong enough to suggest graphs. For instance, more negative attitudes toward fear predicted multicollinearity: Attitude Toward Joy correlated directly with less interest in viewing fear-evoking photographs, and more pos- DES Enjoy (r ϭ .33, p ϭ .001); Attitude Toward Anger correlated itive attitudes toward joy predicted more interest in viewing joy- evoking photographs. Specific ATE subscales also predicted interest in viewing other 3 In presenting regression analyses, we use the language of regression types of emotion photographs. For instance, more negative atti- and denote independent variables as predictor variables. The use of the tudes toward disgust predicted less interest in viewing disgust- term predictor is not intended to imply causation. ATTITUDES TOWARD EMOTIONS 1339

Table 3 Correlations of ATE Subscales With Interest in Viewing Emotional Pictures, Study 2

Variable ATD ATF ATS ATJ ATA D-Int S-Int F-Int J-Int N-Int

ATD — — ءءءATF .48 — ءءء27. ءATS .15 — Ϫ.13 Ϫ.13 ءءATJ Ϫ.20 — Ϫ.12 07. ءءء29. ءءءATA .36 — Ϫ.10 .09 14. ءءء29. ءءءD-Int .39 — ءءءϪ.11 .04 .51 ءء23. ءء21. ءءءS-Int .32 — ءءء47. ءءء45. 09. ءϪ.18 13. ءءء30. ءءءF-Int .31 — ءϪ.13 Ϫ.17 ءءءϪ.10 Ϫ.26 ءϪ.05 .19 ءϪ.15 ءءJ-Int Ϫ.20 — ءءءϪ.05 Ϫ.09 Ϫ.08 Ϫ.02 Ϫ.04 .27 ءN-Int Ϫ.12 Ϫ.13 Ϫ.17

Note. ATE ϭ Attitudes Toward Emotion Scale; ATD ϭ Attitude Toward Disgust; ATF ϭ Attitude Toward Fear; ATS ϭ Attitude Toward Sadness; ATJ ϭ Attitude Toward Joy; ATA ϭ Attitude Toward Anger; D-Int ϭ interest in viewing disgust photographs; S-Int ϭ interest in viewing sad photographs; F-Int ϭ interest in viewing fear photographs; J-Int ϭ interest in viewing joy photographs; N-Int ϭ interest in viewing neutral photographs. .p Ͻ .001 ءءء .p Ͻ .01 ءء .p Ͻ .05 ء directly with DES Anger (r ϭ .28, p ϭ .001); Attitude Toward the other attitude toward a specific emotion. In addition to pre- Sadness correlated directly with DES Sadness (r ϭ .23, p ϭ .001); dicting interest in viewing disgust stimuli, the attitude toward Attitude Toward Disgust was not correlated with DES Disgust disgust predicted interest in viewing fear-evoking and sadness- (r ϭ .13, p ϭ .08); and Attitude Toward Fear was not correlated evoking stimuli after controlling for attitude toward fear and with DES Fear (r ϭ –.04, p ϭ .61). Lack of multicollinearity was attitude toward sadness, respectively. We are unsure why attitude also supported by tolerance tests and small changes in estimated toward disgust stood out in this particular way, but taken together regression coefficients when the predictor variables were added. with the overall pattern of results, these results supported the Results indicated that participants’ interest in viewing fear stim- general predictive validity of the ATE subscales and theoretical uli was predicted by their attitude toward fear (␤ϭ.30, p Ͻ .001), construct. In addition, these results suggest that attitudes toward but not trait experience of fear (␤ϭ–.03, p ϭ .65). Interest in emotions may predict the kind of activities in which individuals viewing disgust stimuli was predicted most strongly by attitude engage, though future research will be needed to test this idea. toward disgust (␤ϭ.36, p Ͻ .001) and independently by trait disgust, albeit more weakly (␤ϭ.13, p ϭ .05). Interest in viewing Study 3 sadness stimuli was predicted by attitude toward sadness (␤ϭ.22, p Ͻ .001), but not trait sadness (␤ϭ.07, p ϭ .35). Interest in Study 3 was designed to assess the relationships between ATE viewing joy stimuli was predicted by attitude toward joy (␤ϭ.15, subscales and trait measures of specific emotions. We expected p ϭ .04) and independently by trait enjoyment (␤ϭ.16, p ϭ .02). that the trait affect measures would correlate moderately with ATE Means for ATE subscales. For Study 2, the means and subscales but that the direction of correlation between ATE sub- standard deviations for each ATE subscale were as follows: Anger, scales and trait affect would depend on the motivational direction M ϭ 1.91, SD ϭ 0.64; Disgust, M ϭ 1.78, SD ϭ 0.73; Fear, M ϭ of the specific emotion. Approach-oriented emotions are predicted 2.27, SD ϭ 0.83; Sadness, M ϭ 2.57, SD ϭ 0.58; and Joy, M ϭ to be directly correlated with liking for the specific emotion. Thus, 4.79, SD ϭ 0.34. liking for anger should be associated directly with trait anger, and liking for joy should be associated directly with trait positive Discussion affect. In contrast, withdrawal-oriented emotions are predicted to be inversely correlated with liking for the specific emotion. Thus, In Study 2, attitudes toward specific emotions predicted interest liking for fear should be associated inversely with trait fear, and in viewing stimuli that evoke that specific emotion. Specifically, liking for disgust should be associated inversely with trait disgust. attitudes toward fear, disgust, sadness, and joy predicted interest in For sadness, we did not have a definite prediction because sadness viewing stimuli that evoke fear, disgust, sadness, and joy, respec- is often viewed as a low approach emotion (Carver, 2004; Hen- tively, independent of the effects of trait emotions. Whereas atti- riques & Davidson, 2000), but other times it is viewed as a tudes toward emotions consistently predicted interest in attitude- withdrawal emotion (K. A. Buss et al., 2003). relevant emotional stimuli, trait emotional experience did not. As discussed earlier, we base these predictions on the fact that Overall, these results are consistent with the prediction that spe- both emotional experience and attitudes toward emotion assist in cific attitudes toward emotions would relate to interest in attitude- guiding and energizing behavior. That is, they both assist in relevant emotional stimuli. motivating behavior that corresponds to the emotion itself. Disgust Although the results suggested specificity between attitudes and the dislike of disgust both promote avoidance motivation. toward a particular emotion and interest in viewing stimuli asso- When individuals are disgusted, they are motivated to avoid the ciated with that emotion, there were a few cases of attitudes toward situation causing the disgust. In contrast, anger promotes approach disgust predicting interest in viewing stimuli associated with other motivation (Carver & Harmon-Jones, 2009). When individuals are emotions, even after statistically controlling for the influence of angry, they are motivated to approach the situation or target 1340 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE causing the anger. If they like the experience of anger, they should The Beck Depression Inventory (BDI; Beck, Steer, Ball, & be even more motivated to approach the anger and the situation Ranieri, 1996), a 21-item multiple choice inventory, contains items causing anger. relating to depression symptoms such as hopelessness and irrita- In addition, we included a measure of trait approach and with- bility and cognitions such as guilt or feelings of being punished. It drawal motivation, the Behavioral Inhibition/Behavioral Activa- also includes items assessing physical symptoms such as fatigue tion Scales (BIS/BAS; Carver & White, 1994); they relate to other and lack of interest in sex (Cronbach’s ␣ϭ.91). emotional variables (E. Harmon-Jones, 2003; Peterson, Gable, & Carver and White’s (1994) BIS/BAS questionnaire assesses Harmon-Jones, 2008). We predicted that high levels of behavioral individual differences in behavioral inhibition sensitivity (BIS) activation sensitivity (BAS) would relate to more positive attitudes and behavioral activation sensitivity (BAS). The BIS scale con- toward the approach emotions anger and joy. BAS may also relate tains seven items. The BAS scale consists of three subscales: (a) to more positive attitudes toward the withdrawal emotions fear and Reward Responsiveness, which contains five items that “focus on disgust, because individuals high in BAS seek out sensations or positive responses to the occurrence or anticipation of reward”; (b) exciting situations that may occur along with fear and disgust (e.g., Drive, which contains four items that pertain “to the persistent rollercoaster rides). pursuit of desired goals”; and (c) Fun Seeking, which has four items “reflecting both a desire for new rewards and a willingness Method to approach a potentially rewarding event on the spur of the moment” (Carver & White, 1994, p. 322). The BIS/BAS question- Participants and procedure. Participants were 191 introduc- naire has acceptable reliability and validity, and it relates to puta- tory psychology students (138 women, 53 men) who completed a tive substrates of BIS and BAS (Amodio, Master, Yee, & Taylor, set of questionnaires in exchange for extra course credit. The 2008; E. Harmon-Jones & Allen, 1997). As in past research, the questionnaires were completed in groups of 20–30 participants in three subscales were combined to form the index of BAS (sub- a classroom. After providing informed consent, the students com- scales were intercorrelated, rs Ͼ .28, ps Ͻ .001). Responses were pleted questionnaires, presented in a random order. In addition to measured using a 4-point scale (1 ϭ strongly disagree to 4 ϭ the ATE, participants completed the following questionnaires. strongly agree). Cronbach’s alphas were .81 for BAS Total, .62 for Materials. The Positive and Negative Affect Schedule— BAS Reward Responsiveness, .76 for BAS Fun Seeking, .80 for Expanded (PANAS–X; Watson & Clark, 1991) contains 60 items BAS Drive, and .78 for BIS. designed to assess 11 specific affective traits: fear, sadness, guilt, hostility, shyness, fatigue, surprise, joviality, self-assurance, attentiveness, and serenity. Participants were asked to indicate to what Results extent they had felt each of the feelings “in general, that is, on the average.” All subscales were internally consistent in the present Correlations between scores on the ATE subscales and scores on sample (Cronbach’s ␣Ͼ.75). the other questionnaires are presented in separate tables for ease of Because the PANAS–X does not include a measure of disgust comprehension. Some participants failed to complete all question- (perhaps because individuals do not report experiencing disgust as naires; therefore, degrees of freedom differ depending on the a long-term mood), we administered the widely used Sensitivity to questionnaires analyzed. Disgust Scale (Haidt, McCauley, & Rozin, 1994) and the Attitude Correlations with PANAS–X subscales. Table 4 presents Toward Disgust subscale to a separate sample of participants (n ϭ the correlations of the ATE subscales with the PANAS–X sub- 75). The Sensitivity to Disgust Scale contains 32 items, with half scales. As explained earlier, we expected that the trait affective requiring true–false responses and half being rated along a 3-point experience measures would correlate moderately with ATE subscale as 1 (not disgusting),2(slightly disgusting), or 3 (very scales but that the direction of correlation between ATE subscales disgusting). Example items included “I might be willing to try and trait affect would depend on the motivational direction of the eating monkey meat under some circumstances” (true–false) and specific emotion. “You see someone put ketchup on vanilla ice-cream and eat it” We expected that among the approach-oriented emotions, joy (rated on the 1–3 scale). The scale’s reliability and validity have and anger, liking for the emotion would be associated directly with been established (Haidt et al., 1994). In the current sample, both trait emotion. In contrast, among the withdrawal-oriented emo- versions were reliable (true–false scale: Cronbach’s ␣ϭ.68; tions, fear and disgust, liking for the emotion would be associated ratings scale: Cronbach’s ␣ϭ.81). inversely with trait emotion. We did not have a specific prediction The A. H. Buss and Perry (1992) Aggression Questionnaire regarding sadness, as previous work has viewed it as an approach includes 29 items that form four subscales: (a) Physical Aggres- (Henriques & Davidson, 2000) or withdrawal emotion (K. A. Buss sion, which assesses the frequency of physically aggressive be- et al., 2003). havior; (b) Verbal Aggression, which assesses the frequency of Consistent with hypotheses, individuals who had a relatively verbally aggressive behavior; (c) Anger, which assesses the emo- positive attitude toward anger reported greater trait hostility but tional component of aggression; and (d) Hostility, which assesses lower trait PA, joy, attentiveness, serenity, and surprise than indi- the cognitive component of aggression, described as “feelings of ill viduals who had a relatively negative attitude toward anger. will and injustice” (A. H. Buss & Perry, 1992, p. 457). As in past Attitude toward joy related to many PANAS–X subscales, such research (E. Harmon-Jones & Allen, 1998), anger was the focus of that individuals with a relatively positive attitude toward joy the present study, although results for the other subscales are also reported greater trait positive activation, joy, self-assurance, atten- presented. All subscales were internally consistent in the present tiveness, and serenity but lower trait negative activation, hostility, sample (Cronbach’s ␣Ͼ.78). guilt, sadness, shyness, and fatigue. ATTITUDES TOWARD EMOTIONS 1341

Table 4 Correlations of ATE Subscales With PANAS–X Subscales, Study 3

ATE subscale NA PA FE Hos GU Sad Joy SA AT Shy FA SE SU

Ϫ.05 .10 .08 .10 Ϫ.13 ءءϪ.20 ءءϪ.18 ءء22. ءءϪ.02 .13 .21 ءATS .06 Ϫ.15 Ϫ.05 Ϫ.09 .00 Ϫ.02 .07 ءATD .01 Ϫ.02 .02 .12 .07 Ϫ.08 Ϫ.03 .15 ءϪ.16 ءϪ.14 10. 11. ءءϪ.23 07. ءءϪ.21 14. 08. ءء41. 04. ءءATA .14 Ϫ.16 12. ءء23. ءϪ.17 ءءϪ.20 ءء19. ءء19. ءء38. ءءϪ.32 ءءϪ.25 ءءϪ.09 Ϫ.24 ءء27. ءATJ Ϫ.17 Ϫ.06 .04 .11 Ϫ.04 ءϪ.16 ءءϪ.02 Ϫ.02 Ϫ.05 .25 08. ءATF Ϫ.10 Ϫ.06 Ϫ.14

Note. ATE ϭ Attitudes Toward Emotion Scale; ATS ϭ Attitude Toward Sadness; ATD ϭ Attitude Toward Disgust; ATA ϭ Attitude Toward Anger; ATJ ϭ Attitude Toward Joy; ATF ϭ Attitude Toward Fear; NA ϭ Negative Activation; PA ϭ Positive Activation; FE ϭ Fear; Hos ϭ Hostility; GU ϭ Guilt; Sad ϭ Sadness; Joy ϭ Joviality; SA ϭ Self-Assurance; AT ϭ Attentiveness; Shy ϭ Shyness; FA ϭ Fatigue; SE ϭ Serenity; SU ϭ Surprise. .p Ͻ .01 ءء .p Ͻ .05 ء

Individuals with a relatively positive attitude toward disgust with physical aggression, verbal aggression, BAS Total, BAS Fun reported lower sensitivity to disgust (two versions of disgust scale, Seeking, and BAS Drive. Also, attitude toward fear correlated r ϭ –.39, p Ͻ .001, and r ϭ –.44, p Ͻ .001), and they scored inversely with BIS. higher in self-assurance than individuals with a relatively less Finally, attitude toward sadness was directly correlated with positive attitude toward disgust. depression and hostility. That is, individuals with a relatively Individuals with more positive attitudes toward fear reported positive attitude toward sadness reported feeling more depressed lower trait fear and attentiveness but greater self-assurance than as well as hostile. These patterns of correlations are consistent with individuals with more negative attitudes toward fear. predictions. Finally, individuals with a relatively positive attitude toward sad- Gender differences. A 2 (male, female) ϫ 5 (ATE: sadness, ness reported greater trait sadness and guilt than individuals with a disgust, anger, joy, fear) mixed-factors ANOVA was performed to relatively negative attitude toward sadness. In addition, individuals test whether gender was related to attitudes toward specific emo- who had a relatively positive attitude toward sadness reported lower tions. A main effect for ATE and a Gender ϫ ATE interaction trait positive activation, joy, and self-assurance than individuals who occurred, F(4, 708) ϭ 12.604, p Ͻ .001. The interaction indicated had a relatively negative attitude toward sadness. that women and men did not differ in their attitudes toward anger These results were consistent with predictions. The direction of or joy. However, they did differ on the other attitudes. Women had the correlations depended on the motivational direction of the more positive attitudes toward sadness (M ϭ 3.02, SD ϭ 0.69) emotion: Correlations involving anger and joy attitudes and moods than men did (M ϭ 2.67, SD ϭ 0.66), p Ͻ .01. Men had more were direct, whereas correlations involving disgust and fear atti- positive attitudes toward disgust (M ϭ 2.14, SD ϭ 0.56) and fear tudes and moods were inverse. (M ϭ 2.49, SD ϭ 0.85) than women did (disgust, M ϭ 1.93, SD ϭ Correlations with other emotive traits. Table 5 presents the 0.47; fear, M ϭ 1.92, SD ϭ 0.73), ps Ͻ .05. Multiple regressions correlations of the ATE subscales with other trait emotive scales. were run to examine whether gender and any of the attitudes on Attitude toward anger was directly correlated with physical ag- which the genders differed interacted to predict the affective traits gression, verbal aggression, hostility, anger, depression, and BAS reported above. No significant interactions emerged. Drive. Attitude toward joy was directly correlated with BAS Total, Means for ATE subscales. For Study 3, the means and BAS Reward Responsiveness, and BAS Fun Seeking but inversely standard deviations for each ATE subscale were as follows: Anger, correlated with depression, physical aggression, and hostility. M ϭ 1.98, SD ϭ 0.79; Disgust, M ϭ 1.99, SD ϭ 0.51; Fear, M ϭ Attitude toward disgust was directly correlated with BAS Total 2.10, SD ϭ 0.82; Sadness, M ϭ 2.91, SD ϭ 0.69; and Joy, M ϭ and BAS Fun Seeking. Attitude toward fear correlated directly 4.76, SD ϭ 0.41.

Table 5 Correlations of ATE Subscales With Emotive Traits, Study 3

ATE subscale BDI PAg VAg Hos Ang BIS BAS RR Fun Drive

Ϫ.08 Ϫ.05 .01 Ϫ.12 01. 11. ء17. 06. 01. ءءATS .25 14. ءء23. 09. ءءATD .01 .13 .14 .10 .13 Ϫ.08 .21 ءϪ.06 .12 .06 .08 .14 ءء41. ءء28. ءء29. ءء33. ءءATA .20 05. ءء24. ءء31. ءءϪ.09 .08 .25 ءءϪ.09 Ϫ.21 ءϪ.16 ءءATJ Ϫ.38 ء15. ءءϪ.08 .33 ءء21. ءءϪ.30 14. 07. ءء22. ءءATF Ϫ.03 .33

Note. ATE ϭ Attitudes Toward Emotion Scale; ATS ϭ Attitude Toward Sadness; ATD ϭ Attitude Toward Disgust; ATA ϭ Attitude Toward Anger; ATJ ϭ Attitude Toward Joy; ATF ϭ Attitude Toward Fear; BDI ϭ Beck Depression Inventory; PAg ϭ Buss-Perry Physical Aggression; VAg ϭ Buss-Perry Verbal Aggression; Hos ϭ Buss-Perry Hostility; Ang ϭ Buss-Perry Anger; BIS ϭ Carver-White Behavioral Inhibition Sensitivity; BAS ϭ Carver-White Behavioral Activation Sensitivity; RR ϭ BAS Reward Responsiveness; Fun ϭ BAS Fun Seeking; Drive ϭ BAS Drive. .p Ͻ .01 ءء .p Ͻ .05 ء 1342 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE

Discussion approach-related emotions was directly related to greater liking, whereas trait experience of withdrawal-related emotions was in- As expected, attitude toward anger was directly correlated with versely related to liking. Finding similar results for state emotional trait anger, aggression, hostility, and BAS Drive. These relation- experience would be important because these predictions are ships suggest that individuals high in trait anger and aggression unique to the current perspective and would not be predicted by a have greater liking for angry experiences. The present results perspective that considered only emotional valence without incor- suggest that individuals with a more positive attitude toward anger porating motivational direction. may also score higher on the Beck Depression Inventory, which Similar to the predictions (and results) of Study 3, in Study 4 we measures irritability as well as depression. predicted that individuals high in liking for approach-oriented Attitude toward joy was directly correlated with many positive emotions would experience more of the emotions in response to affects and BAS (Total, Reward Responsiveness, and Fun Seek- joy- and anger-evoking stimuli, whereas individuals high in liking ing) and inversely correlated with many negative affects and for withdrawal-oriented emotions would experience less of the depression, aggression, and hostility. Thus, individuals who like emotions in response to fear- and disgust-evoking stimuli. On the joy relatively more than others report greater approach motivation, basis of the results of Study 3 that suggest a direct correlation but less depression, aggression, and hostility. between attitude toward sadness and trait sadness, we expected Attitude toward disgust was correlated inversely with trait dis- greater liking for sadness to directly predict state sadness in gust, suggesting that those who experience greater disgust have response to sadness-evoking stimuli. less positive attitudes toward disgust. Attitude toward disgust was In addition to assessing relations of the ATE subscales with correlated directly with BAS Total and BAS Fun Seeking, sug- emotional reactivity, we also included measures of ideal affect gesting that individuals high in approach motivation have a more (Tsai et al., 2006) and appropriate affect (Eid & Diener, 2001), two positive attitude toward disgusting experiences. constructs somewhat similar to the ATE. However, as noted above, Attitude toward fear was correlated inversely with trait fear, sug- these constructs and their associated research programs differ from gesting that those who experience greater fear have less positive the present one in that they have focused primarily on cultural attitudes toward fear. Attitude toward fear was also correlated directly differences, whereas the present work focuses on individual dif- with aggression and BAS but inversely correlated with BIS. These ferences that operate within a culture and how these individual correlations suggest that individuals high in BIS have less liking for differences relate to other aspects of emotion. fearful experiences, whereas individuals high in BAS or high in aggression have greater liking for fearful experiences. Attitude toward sadness was directly correlated with sadness, Method depression, and hostility. These relationships suggest that individ- Participants were 131 introductory psychology students. After uals high in trait sadness and depression have greater liking for sad being greeted by the experimenter, participants were told that the experiences. The results also suggest that individuals high in study concerned cognition and personality and that during the hostility have greater liking for sadness, perhaps because hostility experiment, they would complete some personality questionnaires is associated with depression. and a computer task. Participants then provided informed consent. These results are consistent with our prediction that the motiva- At the beginning of the session, participants completed the ideal tional direction of emotions would moderate the relationship between affect measure from the Affect Valuation Index (Tsai et al., 2006), the the intensity of trait emotion and liking for that emotion. Specifically, appropriate affect measure from Eid and Diener (2001), and the ATE greater trait approach-oriented emotions such as joy and anger were Scale. Ideal affect is based on the affective circumplex, or the map- directly related to liking of those emotions, whereas greater trait ping of different emotional states onto dimensions of arousal and withdrawal-oriented emotions such as fear and disgust were inversely valence. Thus, it includes eight subscales, or octants: high-arousal related to liking of those emotions. Sadness is considered approach states (aroused, surprised, astonished), high-arousal positive states oriented by some researchers and withdrawal oriented by others. (enthusiastic, excited, strong, elated), positive states (happy, satisfied, Consistent with results for the other approach-oriented emotions, trait content), low-arousal positive states (calm, at rest, relaxed, peaceful, sadness was directly related to liking for sadness. These results serene), low-arousal states (quiet, still, passive), low-arousal negative provide evidence for the construct of attitudes toward emotions being states (dull, sleepy, sluggish), negative states (sad, lonely, unhappy), separate from the construct of trait emotions, because if attitudes and high-arousal negative states (fearful, hostile, nervous). Partici- tapped the same construct as trait emotions, the correlations between pants were asked to rate “how much they ideally want to feel each attitudes and traits would have all been direct. state on average” (on a scale from 1 ϭ not at all to 5 ϭ an extreme One of the most consistent set of results involving other inventories amount). In the present sample, the subscales were for the most part concerned BAS. That is, BAS related to several attitudes toward internally consistent (Cronbach’s ␣s Ͼ .63). However, internal con- emotions, such that higher levels of BAS were associated with more sistency was lower for the low-arousal positive (.57), low arousal liking of anger, joy, fear, and disgust. These emotions are all moderate (.44), and high-arousal negative (.59) subscales, although these coef- to high in arousal. Compared with individuals low in BAS, individ- ficients were similar to the published alpha levels in Tsai et al. (2006). uals high in BAS may like arousing emotional experiences, perhaps The Eid and Diener measure asks participants “to indicate how because these individuals like excitement and sensations. appropriate or desirable it is to experience certain emotions” on 7-point scales (1 ϭ extremely desirable and appropriate,2ϭ desir- Study 4 able and appropriate,3ϭ slightly desirable and appropriate,4ϭ Study 4 was designed to test whether attitudes toward emotions neutral [neither desirable nor undesirable],5ϭ slightly undesirable would relate to emotional reactivity. In Study 3, trait experience of and inappropriate,6ϭ undesirable and inappropriate, and 7 ϭ ATTITUDES TOWARD EMOTIONS 1343 extremely undesirable and inappropriate). Emotion items included more sadness than anger and fear (p Ͻ .02), but nonsignificantly “joy,” “affection,” “pride,” “contentment,” “anger,” “fear,” “sadness,” different amounts of fear and anger (p Ͼ .70). and “guilt.” The items were not combined in Eid and Diener. In response to pictures intended to evoke fear, a significant Participants then viewed six sets of pictures. Each set consisted effect occurred, F(4, 520) ϭ 98.92, p Ͻ .001. It indicated that fear of one emotion category (neutral, fear, sadness, disgust, joy, and pictures evoked more fear than any other emotion (ps Ͻ .001). anger). Five of the picture sets were those used in Study 3. The Fear pictures evoked more disgust, anger, and sadness than joy anger picture set was designed to evoke anger, with images of (ps Ͻ .001). Fear pictures caused more disgust than anger and burning United States flags and the 9/11 World Trade Center sadness (ps Ͻ .02) and more anger than sadness (p Ͻ .03). attacks (collected from the Internet). The emotional pictures had the intended effects on emotional Pictures were presented for 6 s. The neutral picture set was always experience (even though the negative emotion pictures often presented first, followed by the five other emotional picture sets evoked slight increases in other negative emotions). Next, we presented in a random order. Following each picture set, participants examine relations between attitudes toward emotions and emo- rated how much they felt the following emotions: afraid, fearful, tional reactions to these pictures. angry, mad, sad, down, disgusted, grossed out, happy, and joyful (on Correlations between ATEs and emotional reactivity. 1 ϭ not at all to 7 ϭ extremely scales). Similar emotion words were More approach-oriented emotional experiences are predicted to be averaged together (e.g., angry and mad). associated with more liking of the corresponding emotions, whereas more withdrawal-oriented emotional experiences are predicted to be Results associated with more disliking of the corresponding emotions. Emotional reactivity to the picture manipulation. We first Consistent with these expectations, attitude toward joy corre- tested whether the pictures presented to participants evoked the lated directly with self-reported joy in response to the joy pictures. intended emotion by conducting a 6 (picture set) ϫ 5 (emotion Attitude toward joy did not correlate with any other emotional ratings) repeated-measures ANOVA. It produced a significant reactions, except anger experienced toward the sad pictures. See interaction, F(20, 2580) ϭ 315.29, p Ͻ .001. We probed this Table 7 for all correlations of ATE subscales with emotional interaction by examining the emotion ratings within each picture reactions to pictures. set using repeated-measures ANOVA. As expected, attitude toward anger correlated directly with In response to pictures intended to evoke joy, a significant effect experienced anger in response to the anger pictures. Attitude ϭ Ͻ occurred, F(4, 520) 834.23, p .001. It indicated that joy pictures toward anger also correlated inversely with fear to the anger Ͻ evoked more self-reported joy than any other emotion (ps .001), pictures, inversely with experienced joy to the joy pictures, and and the other emotions did not differ from one another (see Table 6). inversely with experienced anger to the disgust pictures. In response to pictures intended to evoke anger, a significant effect Predictions were supported for the withdrawal-oriented emo- occurred, F(4, 520) ϭ 182.10, p Ͻ .001. It indicated that anger tions as well. Attitude toward disgust correlated inversely with pictures evoked more self-reported anger than any other emotion experienced disgust to the disgust pictures. Attitude toward disgust (ps Ͻ .001). Also, all emotions differed from each other (ps Ͻ .02). also correlated inversely with experienced anger, sadness, and fear In response to pictures intended to evoke sadness, a significant effect occurred, F(4, 520) ϭ 296.18, p Ͻ .001. It indicated that sad to the disgust pictures. In addition, in response to the fear pictures, pictures evoked more self-reported sadness than any other emotion attitudes toward disgust correlated inversely with experienced sad- (ps Ͻ .001). Also, sad pictures caused greater anger, disgust, and ness, disgust, and fear. fear than joy (ps Ͻ .001). Sad pictures caused more fear than Attitude toward fear correlated inversely with experienced fear disgust (p Ͻ .02), but nonsignificantly different amounts of fear to the fear pictures. Attitude toward fear also correlated inversely and anger (p Ͼ .20) and anger and disgust (p Ͼ .20). with experienced anger and disgust to the fear pictures. In response In response to pictures intended to evoke disgust, a significant to the anger pictures, attitude toward fear correlated inversely with effect occurred, F(4, 516) ϭ 227.59, p Ͻ .001. It indicated that experienced sadness and fear. In response to the disgust pictures, disgust pictures evoked more self-reported disgust than any other attitude toward fear correlated inversely with experienced disgust. emotion (ps Ͻ .001). Also, disgust pictures evoked more anger, Attitude toward sadness correlated directly but not significantly sadness, and fear than joy (ps Ͻ .001). Disgust pictures caused with sadness to the sad pictures (r ϭ .13, p ϭ .13; we report exact

Table 6 Mean (SD) Ratings of Experienced Emotion in Response to Emotional Pictures, Study 4

Picture type Joy Anger Sad Disgust Fear

Neutral pictures 2.37 (1.35) 1.21 (0.57) 1.36 (0.70) 1.17 (0.48) 1.30 (0.64) Joy pictures 5.16 (1.36) 1.12 (0.65) 1.10 (0.30) 1.12 (0.35) 1.23 (0.52) Angry pictures 1.17 (0.58) 4.96 (1.80) 3.28 (1.83) 2.89 (1.64) 2.15 (1.52) Sad pictures 1.30 (0.65) 1.79 (1.09) 4.70 (1.40) 1.66 (1.13) 1.93 (1.29) Disgust pictures 1.23 (0.61) 2.74 (1.74) 3.11 (1.60) 5.74 (1.57) 2.68 (1.85) Fear pictures 1.29 (0.64) 2.52 (1.68) 2.03 (1.29) 2.85 (1.87) 4.05 (1.78)

Note. The boldfaced means (and SDs) are the targeted experienced emotion for the emotional pictures. 1344 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE

Table 7 icant predictor (␤ϭ.13, p ϭ .13). For disgust pictures, the only Correlations of ATE Subscales With Emotional Reactions to significant predictor was attitude toward disgust (␤ϭ–.33, p ϭ Pictures, Study 4 .001). For fear pictures, the only significant predictor was fear ratings to neutral pictures (␤ϭ.21, p ϭ .02). Thus, all ATEs Picture type—rating ATA ATJ ATS ATD ATF showed unique relations with targeted emotional reactions to the Neut-Joy Ϫ.13 Ϫ.03 Ϫ.00 .10 Ϫ.03 targeted stimuli except for attitude toward fear. This latter result Neut-Anger Ϫ.11 Ϫ.00 Ϫ.09 Ϫ.08 Ϫ.09 may have occurred because both attitude toward fear and attitude Ϫ Ϫ Ϫ Ϫ Neut-Sad .08 .16 .01 .02 .01 toward disgust predict fear responses, and these two ATEs are Neut-Disgust Ϫ.13 Ϫ.11 Ϫ.13 Ϫ.06 Ϫ.13 Neut-Fear Ϫ.15 .06 .07 Ϫ.07 Ϫ.15 highly correlated, perhaps because they are withdrawal related, Anger-Joy Ϫ.13 .03 Ϫ.04 Ϫ.08 Ϫ.07 negative, and high in arousal. It is important to note that although Ϫ.14 Ϫ.06 .01 disgust and anger are high-arousal, negative emotions that occupy 05. ءAnger-Anger .18 ءءء ء Anger-Sad Ϫ.01 .11 .16 Ϫ.19 Ϫ.29 the same space in Valence ϫ Arousal dimensions, attitude toward Anger-Disgust .11 Ϫ.04 Ϫ.02 Ϫ.17 Ϫ.16 disgust was inversely correlated with disgust reactions, but attitude ءء ءء ء Anger-Fear Ϫ.18 .06 .11 Ϫ.23 Ϫ.25 .toward anger was directly correlated with anger reactions ءء ءءء ء Joy-Joy Ϫ.20 .34 .01 Ϫ.25 Ϫ.13 Joy-Anger .12 .04 Ϫ.08 .14 Ϫ.05 Correlations between ATE and other measures. Joy-Sad Ϫ.03 Ϫ.09 .01 .07 Ϫ.04 Ideal affect. Attitudes toward emotions were moderately cor- Joy-Disgust .03 Ϫ.06 .00 .04 Ϫ.04 related with some of the ideal affect measures from the Affect Val- Joy-Fear Ϫ.06 .10 Ϫ.08 Ϫ.05 Ϫ.01 uation Index (see Table 8). These correlations were, for the most part, Ϫ ء Ϫ Ϫ Sad-Joy .04 .18 .04 .07 .07 in expected directions. For instance, attitudes toward anger, fear, and Sad-Anger .04 .01 .05 .02 Ϫ.05 Sad-Sad Ϫ.07 .11 .13 Ϫ.15 Ϫ.13 disgust were correlated directly with ideal high-arousal negative af- Sad-Disgust Ϫ.01 Ϫ.05 Ϫ.01 .09 Ϫ.01 fect. Attitude toward joy correlated directly with several ideal positive Sad-Fear Ϫ.17 Ϫ.02 Ϫ.03 Ϫ.08 Ϫ.09 affects and inversely with several ideal negative affects. However, Disgust-Joy Ϫ.07 Ϫ.12 .08 .15 Ϫ.03 attitude toward sadness was not correlated with ideal low-arousal Ϫ.11 ءϪ.04 Ϫ.21 14. ءDisgust-Anger Ϫ.21 Ϫ.09 negative affect. Inspection of correlations of this ATE subscale with ءDisgust-Sad Ϫ.04 .05 .12 Ϫ.18 Ϫ.06 individual items from the ideal affect measure revealed that attitude ءءءDisgust-Disgust Ϫ.07 Ϫ.00 Ϫ.08 Ϫ.28 toward sadness correlated significantly and inversely with arousal ءϪ.22 ءDisgust-Fear Ϫ.14 .05 .09 Ϫ.22 Fear-Joy .01 Ϫ.09 .04 .04 Ϫ.11 (r ϭ –.22, p ϭ .01). The correlation of attitude toward sadness and ءء Fear-Anger Ϫ.04 .04 Ϫ.05 Ϫ.05 Ϫ.23 ideal sadness was close to significant (r ϭ .15, p ϭ .08). Fear-Sad Ϫ.05 .05 .02 Ϫ.17 Ϫ.15 Appropriate affect. The correlations of the ATE subscales and ءءϪ.25 ءFear-Disgust Ϫ.16 .11 Ϫ.06 Ϫ.18 .the Eid and Diener (2001) single-item measures were as expected ءϪ.22 ءءFear-Fear Ϫ.13 .19 Ϫ.08 Ϫ.26

Note. Picture type is indicated by the first word in the left-most column; emotion rating is the second word in the same column. ATE ϭ Attitudes Toward Emotion Scale; ATA ϭ Attitude Toward Anger; ATJ ϭ Attitude Toward Joy; ATS ϭ Attitude Toward Sadness; ATD ϭ Attitude Toward Table 8 Disgust; ATF ϭ Attitude Toward Fear. Correlations of ATE Subscales With Ideal Affect and ءءء ءء ء p Ͻ .05. p Ͻ .01. p Ͻ .001. Appropriate Affect Measures, Study 4

Measure ATA ATJ ATS ATD ATF ps relevant to hypotheses; see Gigerenzer, 2004). Attitude toward sadness did correlate directly with sadness to the anger pictures. Ideal affect ءءء For the neutral pictures, no significant correlations emerged High-arousal positive Ϫ.07 .41 Ϫ.10 Ϫ.11 .03 Ϫ.14 Ϫ.17 Ϫ.13 ءءءPositive Ϫ.12 .40 ءء ءء ء -between attitudes toward specific emotions and experienced emo Low-arousal positive Ϫ.19 .25 Ϫ.02 Ϫ.26 Ϫ.12 tional reactions. Low arousal .05 Ϫ.11 .07 .08 Ϫ.00 ء18. ءء27. 02. ءءءϪ.42 ءRegressions controlling for other ATEs. To control for the Low-arousal negative .17 09. 15. 15. ءءءinfluence of other ATEs on the predicted relationship between a Negative .10 Ϫ.36 ء ء ءءء ء given ATE (e.g., attitudes toward anger) and its associated emo- High-arousal negative .20 Ϫ.33 .05 .21 .21 ء19. ءHigh arousal Ϫ.06 .12 Ϫ.17 .19 tional reaction (e.g., reported anger) to the target stimulus (e.g., Appropriate affect Ϫ.13 ءءϪ.15 Ϫ.27 ءءءanger pictures), regression analyses were conducted in which all Joy Ϫ.14 .58 Ϫ.10 ءϪ.14 Ϫ.22 ءءءATEs and the targeted emotional reaction (e.g., reported anger) to Affection Ϫ.10 .38 the neutral stimulus were entered as predictors of the targeted Pride .13 .13 Ϫ.15 Ϫ.03 .13 Contentment .12 .09 .08 Ϫ.16 Ϫ.06 ءء ء ءءء ءءء .(emotion to the targeted stimulus (e.g., anger pictures Anger .47 Ϫ.32 .17 .17 .24 ءءءϪ.09 .09 .12 .32 ءFor anger pictures, the only significant predictor was attitude Fear .22 Ϫ.01 .08 ء22. ءϪ.19 ءtoward anger (␤ϭ.23, p ϭ .01). For joy pictures, attitude toward Sadness .18 Ϫ.07 .11 .02 Ϫ.02 ءjoy was a significant predictor (␤ϭ.25, p ϭ .001); attitude toward Guilt .19 disgust was a significant (inverse) predictor (␤ϭ–.21, p ϭ .03); Note. ATE ϭ Attitudes Toward Emotion Scale; ATA ϭ Attitude Toward and joy ratings to neutral pictures were also a significant predictor Anger; ATJ ϭ Attitude Toward Joy; ATS ϭ Attitude Toward Sadness; (␤ϭ.36, p Ͻ .001). For sad pictures, no predictors were signif- ATD ϭ Attitude Toward Disgust; ATF ϭ Attitude Toward Fear. .p Ͻ .001 ءءء .p Ͻ .01 ءء .p Ͻ .05 ء -icant, but attitude toward sadness remained a direct but nonsignif ATTITUDES TOWARD EMOTIONS 1345

Attitudes toward anger, joy, sadness, and fear were directly correlated Discussion with appropriate anger, joy, sadness, and fear, respectively. Correlations of emotional reactivity with ideal affect and As predicted, attitude toward anger related to more experienced appropriate affect. To reduce the large number of possible anger when anger was the primary emotion evoked, and attitude correlations, we focused on examining only those correlations toward joy related to more experienced joy when joy was evoked. relevant to the predictions for ATE and specific emotional reac- As predicted, attitude toward fear related to less experienced fear tions to the targeted emotional pictures (e.g., reported anger to when fear was evoked, and attitude toward disgust related to less experienced disgust when disgust was evoked. anger pictures). See Table 9 for results. For the Tsai et al. (2006) In addition to these tests, Study 4 also examined the relations ideal affect measure, no significant correlations emerged between between ATE subscales and the ideal affect measure of Tsai et al. ideal affect and experienced targeted emotions. For instance, ideal (2006) and the appropriate affect measure of Eid and Diener (2001). high-arousal negative affect did not correlate with experienced fear For the most part, ATE subscales related in expected ways with ideal to fear pictures. We suspect that the lack of correlations involving affects and appropriate affects, but the correlations were modest. We ideal high-arousal negative affect may have occurred because this suspect some of these correlations were modest because the ideal octant mixes fear and anger. Our results suggest that the attitudes affect measure combines fear and anger into high-arousal negative toward these emotions should not be mixed. Although anger and affect, and our theoretical approach separates fear and anger. Also, fear are both high-arousal, negative emotions, fear is withdrawal the appropriate affect measure includes only one item for each emo- motivated, whereas anger is approach motivated, and motivational tion and may have less reliability than the ATE. direction is important in the relationship between attitudes toward Finally, the ideal affect and appropriate affect measures performed emotions and the experience of emotions. less well than the ATE in predicting experienced emotional reactions For the Eid and Diener (2001) appropriate affect measure, several to emotional stimuli. The ideal affect measure yielded no significant expected correlations emerged. For example, appropriate joy corre- correlations, but the appropriate affect measure yielded two of three lated directly with experienced joy to joy pictures. Appropriate anger expected correlations. The ATE subscales, on the other hand, corre- correlated directly with experienced anger to anger pictures. However, lated in predicted directions with experienced emotional reactions. appropriate fear did not correlate with fear to fear pictures. We did not conduct regressions controlling for appropriate affect on ATE and Study 5 emotional reactivity relationships because ATE and appropriate affect measures are conceptually similar. In Study 5, we examined the relationship between attitude Means for ATE subscales. For Study 4, the means and toward emotion and efforts at emotion regulation. Study 2’s results standard deviations for each ATE subscale were as follows: Anger, could also be interpreted in terms of emotion regulation, but we M ϭ 1.97, SD ϭ 0.81; Disgust, M ϭ 1.89, SD ϭ 0.70; Fear, M ϭ believe that a stronger case for emotion regulation can be made if 2.21, SD ϭ 0.86; Sadness, M ϭ 2.34, SD ϭ 0.71; and Joy, M ϭ individuals are induced to experience emotion and then they are 4.75, SD ϭ 0.39. given an opportunity to regulate that evoked emotion. Individuals may attempt to reduce their experience of negative emotion because they dislike that emotion. However, this assumption has never been tested. Therefore, in the present study, we tested Table 9 whether attitudes toward fear would predict attempts to down- Correlations of Emotional Reactions to Pictures With Ideal regulate fear specifically following the evocation of fear, because Affect and Appropriate Affect Measures, Study 4 fear is an emotion individuals often try to down-regulate. Partic- Measure Anger Joy Sadness Disgust Fear ipants were exposed to a neutral or fear-eliciting film clip and then were given an opportunity to regulate their emotional experience Ideal affect by choosing to expose themselves to something other than fearful ء High-arousal positive .07 .20 .01 Ϫ.02 .09 or other negative stimuli. Gross and Thompson (2007, p. 11) Ϫ Ϫ Positive .03 .12 .02 .01 .11 suggested this as one way to regulate an emotion, when they wrote, Low-arousal positive Ϫ.14 .10 Ϫ.07 Ϫ.07 .04 Low arousal .04 .00 .15 .09 .09 “emotion regulation involves taking actions that make it more (or less) likely that we will end up in a situation we expect will give ءϪ.12 Ϫ.03 Ϫ.19 ءLow-arousal negative Ϫ.01 Ϫ.19 Negative Ϫ.06 Ϫ.08 .10 .03 Ϫ.09 rise to desirable (or undesirable) emotions.” High-arousal negative .13 Ϫ.03 .15 .11 .04 High arousal .12 .15 .10 .02 Ϫ.03 Appropriate affect Method ء19. 10. ءء26. ءءءJoy .04 .34 ءء Affection Ϫ.01 .23 .13 Ϫ.06 .09 Participants were 97 introductory psychology students who par- Ϫ Ϫ Ϫ Pride .10 .06 .04 .07 .12 ticipated in exchange for extra course credit (66% of participants Contentment .06 Ϫ.03 .15 .04 .17 Ϫ.07 .01 Ϫ.03 were women; results were similar across both genders). At least 1 ءءϪ.23 ءAnger .19 Fear .14 Ϫ.14 .03 .10 .02 month prior to the lab study, participants completed the ATE and Sadness .09 Ϫ.00 .02 Ϫ.01 .09 the Differential Emotions Scale (DES–IV; Izard et al., 1993). Guilt .14 .01 .04 .03 .13 Scores were averaged for the Fear subscale to produce a trait fear ␣ϭ Note. The correlations are for specific emotional reactions to the targeted score (Cronbach’s .80). emotional pictures (e.g., reported anger to anger pictures). Participants were run individually and informed that the exper- .p Ͻ .001. iment was testing materials that would be used in future studies ءءء .p Ͻ .01 ءء .p Ͻ .05 ء 1346 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE

After providing informed consent, they watched a short film clip toward fear and interest in viewing fear photographs was signifi- and completed a questionnaire about it. Participants were ran- cantly larger in the fear film condition than in the no-fear film domly assigned to watch either a no-fear video clip (chase scene condition (z ϭ 1.64, p ϭ .05; one-tailed). This effect is consistent from Bullit) or a fear-inducing video clip (hiding in closet scene with the prediction that the evocation of fear would cause individ- from Halloween); the clips were equal in length (5 min) and Bullit uals with more negative attitudes toward fear to be less interested was chosen because it is an arousing film likely to engage attention in exposing themselves to further instances of fear. as much as Halloween. Following the viewing of the film, partic- Regressions controlling for other variables. Within the fear ipants completed a questionnaire that asked them to indicate the film condition, attitude toward fear related to interest in viewing extent to which they experienced certain emotions “right now,” on disgust (r ϭ .38, p Ͻ .01) and sad photographs (r ϭ .30, p Ͻ .03) but scales ranging from 0 (none)to8(the most in my life). Emotion inversely related to interest in viewing neutral photographs (r ϭ –.27, words were “fearful,” “sad,” “happy,” “disgusted,” “afraid,” p Ͻ .05). A simultaneous regression analysis in which attitude toward “down,” “joyful,” and “grossed out.” fear and interest in viewing each type of nonfear photograph predicted Next, responses to the video clip manipulation were assessed interest in viewing fear photographs revealed that attitude toward fear using the measure described in Study 2. That is, participants predicted interest in viewing fear photographs above and beyond viewed five separate sheets of paper, each displaying five color interest in any other type of photograph (␤ϭ.28, p ϭ .01). pictures taken from the IAPS. Prior to giving participants a note- In addition, we tested whether the link between attitude toward fear book containing the five pieces of paper, the experimenter ex- and emotion regulation was driven by fear reactivity (fear to the film). plained that the participant should look over the pictures and To do this, a regression was conducted in which attitude toward fear complete questionnaires about them. The experimenter also ex- and fear reactivity were entered simultaneously to predict interest in plained that the participant should select one of the five picture sets viewing fear pictures. In this regression, only attitude toward fear to be viewed during the remainder of the experiment. That is, they significantly predicted interest in viewing fear pictures (␤ϭ.56, p Ͻ would be able to view only one of the five types of pictures .001). Conceptually replicating Study 4, attitude toward fear was because of time constraints, and they would view 20 pictures of inversely correlated with fear reactivity (r ϭ –.43, p ϭ .001). that picture type and provide ratings of those 20 pictures. Follow- To assess whether this interest in viewing fear stimuli was due to ing the completion of the questionnaires about the five picture sets attitude toward fear or trait fear, another regression analysis was and prior to the expected viewing of the additional 20 pictures, the conducted in which attitude toward fear and trait fear were entered experimenter reentered the participant’s cubicle and began a fun- simultaneously, and interest in viewing fear-evoking photographs was neled debriefing in which the participant was carefully probed for the outcome (within the fear film condition). Results indicated that suspicion and then informed of the hypotheses. Participants never attitude toward fear uniquely predicted interest in viewing fear pho- viewed the 20 pictures; we were interested only in their ratings of tographs (␤ϭ.48, p Ͻ .001), overall R2 ϭ .27, F(2, 53) ϭ 9.61, p Ͻ the picture types given their expectation that their picture ratings .001. The effect of trait fear was not significant (␤ϭ–.10, p ϭ .39). would relate to an actual decision. Planned contrasts. The above results support the hypothesis that individuals with the most negative attitudes toward fear would Results be the most motivated to avoid further fear after viewing the fear film clip, as measured by ratings of desire to view fear-inducing Emotional reactivity to film manipulation. As expected, photographs. After the no-fear film, individuals with the most the fear film evoked more fear than the no-fear film. This effect negative attitudes toward fear would also be motivated to avoid ϫ was revealed in a significant 2 (film clip) between-subjects 8 fear but not as much as the individuals who just viewed the fear ϭ Ͻ (emotion item) within-subject ANOVA, F(7, 665) 7.39, p film. Finally, individuals with less negative attitudes toward fear .001. Compared with the no-fear film, the fear film caused higher would not be as motivated to avoid fear regardless of fear film ϭ ϭ ϭ ratings on the items “fear” (M 1.39, SD 1.34 vs. M 2.32, condition. To further test this specific predicted pattern, a median ϭ ϭ ϭ ϭ SD 1.96) and “afraid” (M 1.32, SD 1.29 vs. M 2.13, split was performed on attitude toward fear scores. Then, the ϭ SD 2.01). The fear film also caused higher ratings on “dis- following set of contrast coefficients, as recommended by Rosnow ϭ ϭ ϭ ϭ gusted” (M 0.85, SD 1.04 vs. M 1.87, SD 1.89) and and Rosenthal (1995),4 was used to test the predicted pattern of ϭ ϭ ϭ ϭ “grossed out” (M 0.34, SD 0.62 vs. M 1.66, SD 2.03), results: high negative attitudes toward fear/fear film condition ps Ͻ .05. The films were not rated differently on “sad,” “down,” “joyful,” or “happy” (ps Ͼ .06). Correlations of attitude toward fear and interest in viewing 4 We tested our predicted pattern of results using planned contrasts, on the fear pictures. Within the no-fear film condition, attitude toward basis of recommendations of Rosenthal and Rosnow (1985), Rosnow and fear was not significantly related to interest in viewing any of the Rosenthal (1995), and Rosenthal, Rosnow, and Rubin (1999). Contrast anal- photographs. However, attitude toward fear was directly, though ysis is designed to address focused, predicted tests of the data. It provides not significantly, related to interest in viewing fear photographs greater statistical power and clearer substantive interpretations of the results. A (r ϭ .21, p ϭ .20). This correlation is of similar direction and significant interaction from an ANOVA can occur even when the results are not in line with predictions. Furthermore, an interaction is a planned compar- magnitude to that found in Study 2 but did not reach significance, ison. And a 2 ϫ 2 interaction is most likely to be significant when the data perhaps because of the smaller sample size in the current study. form a perfect crossover pattern (i.e., an “X” pattern), that is, when an Our central hypothesis concerned participants’ responses after interaction is testing a contrast of 1 Ϫ1 Ϫ1 1. This “X” pattern is clearly not watching the fear film. Within this condition, attitude toward fear predicted by the current work. That is, an “X” pattern would predict that was significantly related to interest in viewing fear photographs individuals who dislike fear the most should demonstrate the greatest approach (r ϭ .51, p Ͻ .001). Importantly, this correlation between attitude toward fear stimuli in the no-fear film condition. ATTITUDES TOWARD EMOTIONS 1347

(–3); high negative attitudes toward fear/no-fear film condition 2.28, SD ϭ 0.88; Sadness, M ϭ 2.61, SD ϭ 0.56; and Joy, M ϭ (–1); low negative attitudes toward fear/fear film condition (2); 4.81, SD ϭ 0.29. low negative attitudes toward fear/no-fear film condition (2). This planned comparison was significant, t(90) ϭ 4.37, p Ͻ .001. Discussion Moreover, the high negative attitudes toward fear/fear film condi- Study 5 revealed that attitudes toward emotion relate to efforts tion differed from all other conditions (ps Ͻ .05). Similar results at emotion regulation, tested here in the context of fear. Specifi- were obtained when the extreme quartiles on attitude toward fear cally, individuals with the greatest dislike of fear were the ones were used. Individuals in the fear film condition with the most most motivated to avoid viewing fear-arousing stimuli after view- negative attitudes toward fear expressed the least interest in viewing a fear-inducing film clip. These results suggest that that indi- ing the fear photographs. viduals attempt to reduce their experience of negative emotions The same set of contrast coefficients was used to test whether because they dislike those emotions. To our knowledge, this is the interest in viewing other emotion picture types conformed to the first direct test of this idea, and consequently, this study extends predicted pattern for attitude toward fear and fear film condition. the implications of attitudes toward emotions to research and None of these comparisons was significant. Interest in viewing the theory on the motivation to down-regulate emotions. neutral pictures, however, did produce a significant but negative t We interpret the current study’s results in terms of emotion value for the planned comparison, t(90) ϭ –2.47, p ϭ .02. This regulation, which is often defined broadly, in that emotion negative t value suggests that the obtained pattern for interest in regulation processes may dampen, intensify, or simply maintain viewing neutral pictures was opposite to the predicted ordering of an emotion. They can be automatic or controlled, conscious or means for interest in viewing fear pictures. As shown in Figure 1, unconscious, and intrinsic or extrinsic processes, and they un- individuals who disliked fear the most and watched the fear film fold over time and thus may influence the latency, rise time, were more interested in viewing neutral pictures than other par- magnitude, duration, and offset of responses in behavioral, ticipants. Knowing they would have to view one of the sets of experiential, or physiological domains (Gross & Thompson, pictures, perhaps these participants replaced their avoidance of the 2007). Following this definition, we suggest that when individ- fear pictures with an interest in viewing the neutral pictures. uals who strongly disliked fear indicated that they did not want Omnibus ANOVA. In response to a reviewer’s request, we to view other fear-provoking images after watching a fear- also conducted a 2 (attitude toward fear) ϫ 2 (fear vs. no-fear inducing film, they were attempting to regulate their fear. film) ϫ 5 (preferences for fear, disgust, sad, neutral, and joy However, our results could be due to mild avoidance or a desire pictures) ANOVA. It produced a significant three-way interaction, to experience different emotions, although these motivations F(4, 360) ϭ 3.55, p ϭ .007. Only interest in viewing fear pictures could also be due to emotion regulation processes. conformed to the predicted pattern tested with the planned comparison. General Discussion Means for ATE subscales. For Study 5, the means and standard deviations for each ATE subscale were as follows: Anger, The present research developed a new approach to the study of M ϭ 1.90, SD ϭ 0.64; Disgust, M ϭ 1.78, SD ϭ 0.75; Fear, M ϭ attitudes toward emotions by examining how they differ among

Figure 1. Interest in viewing emotion-evoking photographs as a function of attitude toward fear and film viewing condition. Neg Att Fear ϭ negative attitude toward fear; Pix ϭ pictures. 1348 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE individuals within a culture and how they predict other important experience. This is possible. However, like Izard (1971), Eid and emotion-related variables. Study 1 suggested that there are specific Diener (2001), and Ekman (2003), we believe that the attitude attitudes toward joy, anger, disgust, fear, and sadness and that toward the emotion may be an intrinsic part of the emotional these distinctions are reliable and valid. Factor analyses supported experience itself. Future research with methods that measure men- the distinction of attitudes for specific emotions above alternative tal processes on the order of milliseconds may be necessary to factor structures. Study 2 revealed that attitudes toward emotions address questions of this sort. predicted emotional situation selection, such that more liking of a Our work is similar to work by Eid and Diener (2001), which particular emotion was correlated with being more likely to want found that different cultures have different norms for the expres- to view stimuli that would evoke that emotion. Together, the first sion and experience of the various emotions. The Eid and Diener two studies validated the basic concept of attitudes toward emotion research is similar and complementary to the current work, al- and demonstrated that the ATE Scale was a reliable measure that though it also differs in some important ways. Results of Study 4, possessed basic predictive validity independent of trait emotions. in particular, revealed some associations between our ATE sub- In general, emotions that are typically thought of as negative were scales and the Eid and Diener one-item measures of appropriate evaluated as negative by individuals. The mean rating for each of emotions. Whereas Eid and Diener focused on cultural differences these emotions was below the midpoint of the scale. However, in emotion norms, the present studies focused on individual dif- individuals showed reliable differences in the degree to which they ferences in emotion attitudes within a culture. Importantly, the evaluated those emotions as negative, and these attitudes corre- present research examined how attitudes toward emotion related to lated with situation selection. emotion situation selection, trait emotions, emotional reactivity, Study 3 tested hypotheses about the relationship between atti- and emotion regulation, things no previous work has done. Future tudes toward emotion and trait emotions. Consistent with hypoth- work should incorporate these two perspectives by testing whether eses, attitudes toward emotions correlated with trait emotions, but cultural differences in attitudes or norms influence emotion regu- the direction of these correlations depended on the motivational lation, emotion reactivity, or situation selection. direction of the emotion. That is, liking of approach-oriented Another related concept is affect valuation theory’s (Tsai et al., emotions, such as anger and joy, was directly associated with more 2006) ideal affect, which assesses the affective states that people of the trait emotion, whereas liking of withdrawal-oriented emo- value and would ideally like to feel. According to affect valuation tions, such as fear and disgust, was inversely associated with more theory, ideal affect primarily involves different positive states, of the trait emotion. Study 4 followed up on Study 3 by examining whereas the current work focuses primarily on different negative the relationships between attitudes toward emotions and the inten- states. Moreover, the present work differs from Tsai et al.’s (2006) sity of reactions to emotional stimuli. It revealed that liking for the conceptualization because it focuses on liking for specific emo- approach emotions joy and anger was correlated with greater tions, not ideal states. Also, our conceptualization emphasizes a experienced joy and anger to stimuli that evoke these emotions, specific emotion perspective that is interested in differences be- whereas liking for the withdrawal emotions fear and disgust was tween emotions such as fear and anger—emotions that are con- correlated with less experienced fear and disgust to stimuli that sidered similar in the ideal affect framework, given that both are evoke these emotions. high in negative valence and arousal. As revealed in Study 3, A major contribution of the attitudes toward emotion construct attitude toward fear correlates inversely with trait fear, whereas is that it provides a theoretical account for why people may choose attitude toward anger correlates directly with trait anger, suggest- to down-regulate a particular emotion. Our model acknowledges ing that anger and fear should be analyzed separately. Also, Study that emotions organize an organism’s adaptive response to an 4 revealed that ATE subscales were better at predicting emotional evocative event, and thus a negative emotional response may be reactivity than the ideal affect scales, perhaps because ideal affect beneficial even though one holds a negative attitude toward it. The fails to take into account motivational direction. Future research on ATE model suggests that attempts to regulate an emotion are attitudes toward emotions would benefit by following the lead of driven by attitudes toward the emotion, independent of the emotion affect valuation theory in examining attitudes toward specific itself. Supporting this idea, Study 5 revealed that individuals who positive emotions, particularly those that differ in approach moti- evaluated fear as more negative were more likely to avoid fear- vational intensity (Gable & Harmon-Jones, 2008, 2010). arousing stimuli after fear was activated. This suggests that attitudes toward fear moderated the degree to which individuals Limitations engaged in attempts at emotion down-regulation, thus extending work on emotion regulation by showing an implicit assumption of The present studies were cross-sectional correlational studies, emotion regulation—that individuals try to down-regulate emo- and as such the direction of causation cannot be determined. For tions they dislike (Gross & Thompson, 2007)—is valid. Hence, instance, does liking an approach emotion lead to experiencing that these results provide a link between the basic concept of attitudes emotion more, or does experiencing an approach emotion more toward emotion and an interesting new direction for theories of lead to liking that emotion more? Does liking a withdrawal emo- emotion regulation. tion lead to experiencing that emotion less, or does experiencing a withdrawal emotion more lead to liking that emotion less? We Comparison to Other Conceptualizations suspect that both directions of causation occur, but future research is necessary to test these ideas. Following presentations of our work on attitudes toward emo- In the current research, we focused on the traditional list of basic tions, some listeners/readers have suggested that attitudes toward emotions and consequently included only one traditional positive emotions are a meta-emotion construct or secondary emotional emotion—joy. This limits our exploration of attitudes toward ATTITUDES TOWARD EMOTIONS 1349 positive emotions, something affect valuation theory has demon- Ekman, P. (2003). Emotions revealed: Recognizing faces and feelings to strated is important in comparisons of low- versus high-arousal improve communication and emotional life. New York, NY: Holt & Co. positive affect in Eastern versus Western cultures. Future research Ellsworth, P. C. (1994). Sense, culture, and sensibility. In S. Kitayama & on attitudes toward emotions should incorporate more positive H. R. Markus (Eds.), Emotion and culture: Empirical studies of mutual emotions. influence (pp. 23–50). Washington, DC: American Psychological Asso- ciation. doi:10.1037/10152-001 Ford, B. Q., Tamir, M., Brunye´, T. T., Shirer, W. R., Mahoney, C. R., & Conclusions Taylor, H. A. (2010). Keeping your eyes on the prize: Anger and visual Psychological Science, 21, The present research provided a new conceptualization of indi- attention to threats and rewards. 1098–1105. doi:10.1177/0956797610375450 vidual differences in attitudes toward specific emotions and pre- Frijda, N. H. (1986). The emotions. London, England: Cambridge Univer- sented a measure showing that five of the most researched emo- sity Press. tions load on five separate attitude factors. Attitudes toward Frijda, N. H., & Tcherkassof, A. (1997). Facial expressions as modes of emotions were associated with trait affective experience, and they action readiness. In J. A. Russell & J. M. Ferna´ndez-Dols (Eds.), The predicted emotional reactivity, emotional situation selection, and psychology of facial expression (pp. 78–102). Cambridge, England: emotion down-regulation. By expanding our understanding of Cambridge University Press. doi:10.1017/CB09780511659911.006 attitudes toward emotions, we will expand and clarify our under- Gable, P. A., & Harmon-Jones, E. (2008). Approach-motivated positive standing of emotional valence because the two concepts are so affect reduces breadth of attention. Psychological Science, 19, 476–482. intertwined. In particular, referring to an emotional (or mood) state doi:10.1111/j.1467-9280.2008.02112.x or trait as positive or negative is ambiguous. Researchers should Gable, P. A., & Harmon-Jones, E. (2010). The motivational dimensional explicitly state their definition of emotional valence and whether it model of affect: Implications for breadth of attention, memory, and cogni- refers to the emotion’s consequences, eliciting circumstances, or tive categorization. Cognition & Emotion, 24, 322–337. doi:10.1080/ subjective feel or attitude. Moreover, by clearly delineating the 02699930903378305 concept of emotional valence, researchers should be better posi- Gigerenzer, G. (2004). Mindless statistics. The Journal of Socio-Economics, 33, 587–606. tioned to empirically address what aspect of emotional valence is Gross, J. J., & Thompson, R. A. (2007). Emotion regulation: Conceptual driving an outcome ostensibly caused by the emotion. The results foundations. In J. J. Gross (Ed.), Handbook of emotion regulation (pp. of Study 5 provided initial evidence in line with this idea: Indi- 3–24). New York, NY: Guilford Press. viduals may be motivated to down-regulate certain negative emo- Haidt, J., McCauley, C., & Rozin, P. (1994). Individual differences in tions because of their dislike for the negative feeling state. sensitivity to disgust: A scale sampling seven domains of disgust elici- tors. Personality and Individual Differences, 16, 701–713. doi:10.1016/ References 0191-8869(94)90212-7 Harmon-Jones, C., Schmeichel, B. J., Mennitt, E., & Harmon-Jones, E. Amodio, D. M., Master, S. L., Yee, C. M., & Taylor, S. E. (2008). (2011). The expression of determination: Similarities between anger and Neurocognitive components of the behavioral inhibition and activation approach-related positive affect. Journal of Personality and Social Psy- systems: Implications for theories of self-regulation. Psychophysiology, chology, 100, 172–181. doi:10.1037/a0020966 45, 11–19. Harmon-Jones, E. (2003). Anger and the behavioural approach system. Beck, A. T., Steer, R. A., Ball, R., & Ranieri, W. F. (1996). Comparison Personality and Individual Differences, 35, 995–1005. doi:10.1016/ of Beck Depression Inventories–IA and –II in psychiatric outpatients. S0191-8869(02)00313-6 Journal of Personality Assessment, 67, 588–597. doi:10.1207/ Harmon-Jones, E. (2004). On the relationship of anterior brain activity and s15327752jpa6703_13 anger: Examining the role of attitude toward anger. Cognition & Emo- Buss, A. H., & Perry, M. (1992). The Aggression Questionnaire. Journal tion, 18, 337–361. doi:10.1080/02699930341000059 of Personality and Social Psychology, 63, 452–459. doi:10.1037/0022- Harmon-Jones, E., & Allen, J. J. B. (1997). Behavioral activation sensi- 3514.63.3.452 tivity and resting frontal EEG asymmetry: Covariation of putative indi- Buss, K. A., Schumacher, J. R. M., Dolski, I., Kalin, N. H., Goldsmith, cators related to risk for mood disorders. Journal of Abnormal Psychol- H. H., & Davidson, R. J. (2003). Right frontal brain activity, cortisol, ogy, 106, 159–163. doi:10.1037/0021-843X.106.1.159 and withdrawal behavior in 6-month-old infants. Behavioral Neurosci- Harmon-Jones, E., & Allen, J. J. B. (1998). Anger and prefrontal brain ence, 117, 11–20. doi:10.1037/0735-7044.117.1.11 activity: EEG asymmetry consistent with approach motivation despite Cacioppo, J. T., Priester, J. R., & Berntson, G. G. (1993). Rudimentary negative affective valence. Journal of Personality and Social Psychol- determinants of attitudes: II. Arm flexion and extension have differential effects on attitudes. Journal of Personality and Social Psychology, 65, ogy, 74, 1310–1316. doi:10.1037/0022-3514.74.5.1310 5–17. doi:10.1037/0022-3514.65.1.5 Harmon-Jones, E., & Peterson, C. K. (2008). Effect of trait and state Carver, C. S. (2004). Negative affects deriving from the behavioral ap- approach motivation on aggressive inclinations. Journal of Research in proach system. Emotion, 4, 3–22. doi:10.1037/1528-3542.4.1.3 Personality, 42, 1381–1385. doi:10.1016/j.jrp.2008.05.001 Carver, C. S., & Harmon-Jones, E. (2009). Anger is an approach-related Henriques, J. B., & Davidson, R. J. (2000). Decreased responsiveness to affect: Evidence and implications. Psychological Bulletin, 135, 183– reward in depression. Cognition & Emotion, 14, 711–724. doi:10.1080/ 204. doi:10.1037/a0013965 02699930050117684 Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral Higgins, E. T., Shah, J., & Friedman, R. (1997). Emotional responses to activation, and affective responses to impending reward and punishment: goal attainment: Strength of regulatory focus as moderator. Journal of The BIS/BAS Scales. Journal of Personality and Social Psychology, 67, Personality and Social Psychology, 72, 515–525. doi:10.1037/0022- 319–333. doi:10.1037/0022-3514.67.2.319 3514.72.3.515 Eid, M., & Diener, E. (2001). Norms for experiencing emotions in different Hopwood, C. J., & Donnellan, M. B. (2010). How should the internal cultures: Inter- and intranational differences. Journal of Personality and structure of personality inventories be evaluated? Personality and Social Social Psychology, 81, 869–885. doi:10.1037/0022-3514.81.5.869 Psychology Review, 14, 332–346. doi:10.1177/1088868310361240 1350 HARMON-JONES, HARMON-JONES, AMODIO, AND GABLE

Izard, C. E. (1971). The face of emotion. New York, NY: Appleton- comparisons in the analysis of variance. New York, NY: Cambridge Century-Crofts. University Press. Izard, C. E. (2009). Emotion theory and research: Highlights, unanswered Rosenthal, R., Rosnow, R. L., & Rubin, D. B. (1999). Contrasts and effect questions, and emerging issues. Annual Review of Psychology, 60, 1–25. sizes in behavioral research: A correlational approach. New York, NY: doi:10.1146/annurev.psych.60.110707.163539 Cambridge University Press. Izard, C. E., Libero, D. Z., Putnam, P., & Haynes, O. M. (1993). Stability Rosnow, R. L., & Rosenthal, R. (1995). “Some things you learn aren’t so”: of emotion experiences and their relations to traits of personality. Jour- Cohen’s paradox, Asch’s paradigm, and the interpretation of interaction. nal of Personality and Social Psychology, 64, 847–860. doi:10.1037/ Psychological Science, 6, 3–9. doi:10.1111/j.1467-9280.1995.tb00297.x 0022-3514.64.5.847 Schwartz, G. E., Fair, P. L., Salt, P., Mandel, M. R., & Klerman, G. L. Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1990). Emotion, attention, (1976, April 30). Facial muscle patterning to affective imagery in and the startle reflex. Psychological Review, 97, 377–395. doi:10.1037/ depressed and nondepressed subjects. Science, 192, 489–491. doi: 0033-295X.97.3.377 10.1126/science.1257786 Lang, P. J., Bradley, M. M., & Cuthbert, B. N. (1999). International Sommers, S. (1984). Adults evaluating their emotions: A cross-cultural Affective Picture System (IAPS): Technical manual and affective ratings. perspective. In C. Z. Malatesta & C. E. Izard (Eds.), Emotion in adult Gainesville, FL: University of Florida, Center for Research in Psycho- development (pp. 319–338). Beverly Hills, CA: Sage. physiology. Tomkins, S. S. (1962). Affect imagery consciousness: Vol. I. The positive Lazarus, R. S. (1991). Emotion and adaptation. New York, NY: Oxford affects. London, England: Tavistock. University Press. Tomkins, S. S. (1963). Affect imagery consciousness: Vol. II. The negative Lerner, J. S., & Keltner, D. (2001). Fear, anger, and risk. Journal of affects. New York, NY: Springer. Personality and Social Psychology, 81, 146–159. doi:10.1037/0022- Tsai, J. L. (2007). Ideal affect: Cultural causes and behavioral conse- 3514.81.1.146 quences. Perspectives on Psychological Science, 2, 242–259. doi: Mendes, W. B., Major, B., McCoy, S., & Blascovich, J. (2008). How 10.1111/j.1745-6916.2007.00043.x attributional ambiguity shapes physiological and emotional responses to Tsai, J. L., Knutson, B., & Fung, H. H. (2006). Cultural variation in affect social rejection and acceptance. Journal of Personality and Social Psy- valuation. Journal of Personality and Social Psychology, 90, 288–307. chology, 94, 278–291. doi:10.1037/0022-3514.94.2.278 doi:10.1037/0022-3514.90.2.288 Olson, J. M., & Zanna, M. P. (1993). Attitudes and attitude change. Annual Watson, D., & Clark, L. A. (1991). The PANAS–X: Preliminary manual for the Positive and Negative Affect Schedule—Expanded Form. Unpub- Review of Psychology, 44, 117–154. doi:10.1146/annurev.ps.44 lished manuscript. .020193.001001 Peterson, C. K., Gable, P., & Harmon-Jones, E. (2008). Asymmetrical frontal ERPs, emotion, and behavioral approach/inhibition sensitivity. Received September 15, 2010 Social Neuroscience, 3, 113–124. doi:10.1080/17470910701612736 Revision received May 23, 2011 Rosenthal, R., & Rosnow, R. L. (1985). Contrast analysis: Focused Accepted June 28, 2011 Ⅲ